HTTP/2 out of the box in Wildfly 10.1

classic Classic list List threaded Threaded
67 messages Options
1234
Reply | Threaded
Open this post in threaded view
|

Re: HTTP/2 out of the box in Wildfly 10.1

Darran Lofthouse
On 07/06/16 21:55, Stuart Douglas wrote:

>
>
> On Wed, Jun 8, 2016 at 6:51 AM, Jason Greene <[hidden email]
> <mailto:[hidden email]>> wrote:
>
>     So after reviewing this thread and discussing with a few folks, I’d
>     like to propose, for 10.1:
>
>     #1b - Same as the previous #1, we don’t enable TLS for management by
>     default for now, but we additionally include an extra cli script to
>     enable TLS.
>
>
> We would leave the cert generation bit in the security realm, but just
> don't enable the HTTPS interface. That way all that is required is for
> the user to add the https="managements-https" attribute.

+1 It is a minimal config change then to use it.


>
> Stuart
>
>
>
>     For 11 I think we should move to TLS by default, perhaps with a
>     configurable URL policy on redirects, and address the incongruence
>     with upgrade over app.
>
>     I think its likely reasonable to redirect by default for 11, but we
>     can hash that out further. One nice thing I had forgotten about is
>     that the JDK will prompt for you to add unknown certs, and this all
>     works with the CLI[1]. So it’s really only non-interactive clients
>     we have to worry about, and that sounds like a reasonable burden for
>     upgrade.
>
>     [1]
>
>     [disconnected /] connect
>     Unable to connect due to unrecognised server certificate
>     Subject    - CN=foo,OU=foo,L=Madison,ST=WI,C=US
>     Issuer     - CN=myServer, OU=test, L=Madison, ST=WI, C=US
>     Valid From - Tue Jun 07 15:22:06 CDT 2016
>     Valid To   - Thu Jun 07 15:22:06 CDT 2018
>     MD5 : cd:68:be:0b:e0:c0:1c:63:d5:2a:85:c8:d1:9d:e7:7d
>     SHA1 : ae:f8:35:fd:09:c9:b3:08:05:59:a6:40:5e:ac:6e:e8:ce:85:72:4b
>
>     Accept certificate? [N]o, [T]emporarily, [P]ermenantly : t
>
>
>
>>     On Jun 7, 2016, at 6:24 AM, Jason T. Greene
>>     <[hidden email] <mailto:[hidden email]>> wrote:
>>
>>     Long term I think we want management using TLS, but that can of
>>     course come in phases. Assuming 2) is one of those phases to come
>>     (either now or later), a following step is that the CLI, and
>>     really any remoting client, should prefer TLS with a defaulted
>>     trust store location that points to the keystore.
>>
>>     With 2) if we have the default of the attribute that forces
>>     redirect be true, and our default config be false, then someone
>>     that carries over their old config would not have a potential
>>     security weakness. If they have a CLI script that adds the https
>>     port, it will fail, hopefully sending a signal to look. Although,
>>     the user might just assume that oh it's there, I don't have to do
>>     anything.
>>
>>     Another interesting thing about 2 is that IIRC we have conflicting
>>     behavior between the app port which doesn't force upgrade and the
>>     management port which does.
>>
>>     So my preference is 2, because at some point we have to do it
>>     anyway, and if we have TLS out of the box might as well use it.
>>
>>     On Jun 6, 2016, at 10:48 PM, Stuart Douglas
>>     <[hidden email] <mailto:[hidden email]>>
>>     wrote:
>>
>>>     So while implementing this I have noticed a potential problem
>>>     that it would be good to get some feedback on.
>>>
>>>     If the management interface has SSL by default then the HTTP
>>>     interface will always redirect to the HTTPS interface. This
>>>     effectively breaks the management API, as clients such as the
>>>     CLI, Arquillian etc will be redirected to HTTPS, and then reject
>>>     the self signed certificate (as they should).
>>>
>>>     I am not sure what to do about this, these are the options as I
>>>     see them:
>>>
>>>     1) Don't enable SSL for the management interface (just for the
>>>     Undertow subsystem). The management interface can still use this
>>>     auto-generation capability, it just won't be enable by default
>>>     (we could even leave the cert in the security domain, but just
>>>     not enable the https interface).
>>>
>>>     2) Disable automatic redirects for HTTP upgrade requests
>>>     (potentially controlled by an attribute). This will allow the CLI
>>>     etc to work, but at the price of potentially reducing security,
>>>     as some connections that would have previously been redirected to
>>>     use HTTPS will no longer do this.
>>>
>>>     3) Enable it by default and leave it broken. We can setup some
>>>     kind of automatic trust store thing so the local CLI works, and
>>>     can get our test suite to work with Arquillian in a similar
>>>     manner. Personally I think this is a terrible idea, but I am
>>>     including it for completeness.
>>>
>>>     Personally I think we should go for 1). Given that this is
>>>     supposed to be about developer usability I don't think having
>>>     management also use SSL as being that important.
>>>
>>>     Stuart
>>>
>>>     On Mon, Jun 6, 2016 at 10:24 PM, Jason T. Greene
>>>     <[hidden email] <mailto:[hidden email]>> wrote:
>>>
>>>         Awesome! Another idea I had on how we could get away with it
>>>         being in server boot, is to have a pre-boot first time setup
>>>         task, either launched from the shell/batch scripts or as a
>>>         special pre-step before the AS module loads. We could then
>>>         report boot time as the time AFTER first time installation
>>>         tasks have completed, which I think is fair because the
>>>         server hasn't yet been started.
>>>
>>>         On Jun 5, 2016, at 11:53 PM, Stuart Douglas
>>>         <[hidden email]
>>>         <mailto:[hidden email]>> wrote:
>>>
>>>>         I have some initial work on this at:
>>>>         https://github.com/stuartwdouglas/wildfly-core/tree/WFCORE-1576
>>>>
>>>>         If you go to https://localhost:9993
>>>>         <https://localhost:9993/> it will generate the certificate
>>>>         (although all that will be served is a 404 page as the
>>>>         console is not installed).
>>>>
>>>>         Stuart
>>>>
>>>>         On Mon, Jun 6, 2016 at 12:46 PM, Stuart Douglas
>>>>         <[hidden email]
>>>>         <mailto:[hidden email]>> wrote:
>>>>
>>>>             I think that would actually end up being more complex.
>>>>
>>>>             Stuart
>>>>
>>>>             On Mon, Jun 6, 2016 at 12:45 PM, Jason T. Greene
>>>>             <[hidden email]
>>>>             <mailto:[hidden email]>> wrote:
>>>>
>>>>                 Another option could be a post boot task. So it's
>>>>                 still eager but don't block completed start. We'd
>>>>                 still need to block Tls ports though. So maybe this
>>>>                 does not help
>>>>
>>>>                 On Jun 5, 2016, at 9:31 PM, Stuart Douglas
>>>>                 <[hidden email]
>>>>                 <mailto:[hidden email]>> wrote:
>>>>
>>>>>                 2048 bits adds close to a second to first boot on
>>>>>                 my machine (obviously subsequent boots are
>>>>>                 unaffected).
>>>>>
>>>>>                 This is probably a bit much, I will work on getting
>>>>>                 a POC for the lazy loading approach implemented.
>>>>>
>>>>>                 Stuart
>>>>>
>>>>>                 On Mon, Jun 6, 2016 at 12:27 PM, Jason T. Greene
>>>>>                 <[hidden email]
>>>>>                 <mailto:[hidden email]>> wrote:
>>>>>
>>>>>                     We should really be generating 2048 bit keys.
>>>>>
>>>>>                     I don't like adding to our boot time, we have
>>>>>                     already seen it grow and this would be yet
>>>>>                     another case.
>>>>>
>>>>>                     On Jun 5, 2016, at 8:57 PM, Stuart Douglas
>>>>>                     <[hidden email]
>>>>>                     <mailto:[hidden email]>> wrote:
>>>>>
>>>>>>                     So I just did up a very quick prototype that
>>>>>>                     generates self signed certificates on startup
>>>>>>                     and it looks like the difference in startup
>>>>>>                     time is negligible (at least when generating
>>>>>>                     1024 bit RSA keys). Even if the difference is
>>>>>>                     measurable it only affects the very first startup.
>>>>>>
>>>>>>                     I think that in order to simplify the
>>>>>>                     implementation of this it may be better to
>>>>>>                     simply generate the key of first startup,
>>>>>>                     instead of attempting to do it lazily.
>>>>>>
>>>>>>                     Stuart
>>>>>>
>>>>>>                     On Sat, Jun 4, 2016 at 12:09 AM, Jason T.
>>>>>>                     Greene <[hidden email]
>>>>>>                     <mailto:[hidden email]>> wrote:
>>>>>>
>>>>>>
>>>>>>>                             What will be default keysize? It has
>>>>>>>                             to be probably choosen to work also
>>>>>>>                             without "Java Cryptography Extension
>>>>>>>                             (JCE) Unlimited Strength Jurisdiction
>>>>>>>                             Policy"
>>>>>>>
>>>>>>>
>>>>>>>                         Probably the largest that is supported
>>>>>>>                         without JCE. It does not matter that
>>>>>>>                         much, self signed certs are inherently
>>>>>>>                         insecure, this is a developer usability
>>>>>>>                         feature, not something that can be used
>>>>>>>                         in production.
>>>>>>
>>>>>>                         IIRC there is actually no limit on RSA key
>>>>>>                         size, it's only symmetric algs that are
>>>>>>                         limited, so we could use a standard 2048
>>>>>>                         bit key without issue.
>>>>>>
>>>>>>
>>>>>>>
>>>>>>>                         Stuart
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>                             On Thu, Jun 2, 2016 at 10:01 PM,
>>>>>>>                             Stuart Douglas
>>>>>>>                             <[hidden email]
>>>>>>>                             <mailto:[hidden email]>>
>>>>>>>                             wrote:
>>>>>>>
>>>>>>>                                 So I guess we should talk about
>>>>>>>                                 how this should actually work.
>>>>>>>
>>>>>>>                                 In terms of auto generating the
>>>>>>>                                 key I was thinking we would need
>>>>>>>                                 to add a new attribute to the
>>>>>>>                                 'keystore' element under the
>>>>>>>                                 security realm, something like
>>>>>>>                                 'auto-generate-cert-host="localhost"'.
>>>>>>>                                 I am not sure what other options
>>>>>>>                                 we would need, or how
>>>>>>>                                 configurable we should make it,
>>>>>>>                                 but as this is for
>>>>>>>                                 testing/development purposes I
>>>>>>>                                 don't think we need to expose
>>>>>>>                                 full control over the certificate
>>>>>>>                                 generation process.
>>>>>>>
>>>>>>>                                 In terms of the implementation we
>>>>>>>                                 could just implement an
>>>>>>>                                 SSLContext wrapper, that can do
>>>>>>>                                 the generation and then create a
>>>>>>>                                 'real' SSLContext the first time
>>>>>>>                                 it is asked to create and SSLEngine.
>>>>>>>
>>>>>>>                                 Stuart
>>>>>>>
>>>>>>>                                 On Fri, Jun 3, 2016 at 3:19 AM,
>>>>>>>                                 Jason Greene
>>>>>>>                                 <[hidden email]
>>>>>>>                                 <mailto:[hidden email]>>
>>>>>>>                                 wrote:
>>>>>>>
>>>>>>>
>>>>>>>                                     > On Jun 2, 2016, at 11:29 AM, Harold Campbell <[hidden email]
>>>>>>>                                     <mailto:[hidden email]>> wrote:
>>>>>>>                                     >
>>>>>>>                                     > On Thu, 2016-06-02 at 09:22 +1000, Stuart Douglas wrote:
>>>>>>>                                     >> Hi All,
>>>>>>>                                     >>
>>>>>>>                                     >> I would like to propose that we add support for HTTP/2 out of the box
>>>>>>>                                     >> in Wildfly 10.1.
>>>>>>>                                     >>
>>>>>>>                                     >
>>>>>>>                                     > This lowly user desperately wants a release containing the fix to WFLY-
>>>>>>>                                     > 6283 sooner rather than later. I'm sure other people have other pet
>>>>>>>                                     > bugs awaiting release.
>>>>>>>                                     >
>>>>>>>                                     > I have no opinion on HTTP/2 being added other than to ask that pent up
>>>>>>>                                     > bug fixes be kept in mind.
>>>>>>>
>>>>>>>
>>>>>>>                                     Hi Harold,
>>>>>>>
>>>>>>>                                     That fix is already in
>>>>>>>                                     master, so it will be
>>>>>>>                                     included in 10.1.
>>>>>>>
>>>>>>>                                     --
>>>>>>>                                     Jason T. Greene
>>>>>>>                                     WildFly Lead / JBoss EAP
>>>>>>>                                     Platform Architect
>>>>>>>                                     JBoss, a division of Red Hat
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>                                 _______________________________________________
>>>>>>>                                 wildfly-dev mailing list
>>>>>>>                                 [hidden email]
>>>>>>>                                 <mailto:[hidden email]>
>>>>>>>                                 https://lists.jboss.org/mailman/listinfo/wildfly-dev
>>>>>>>
>>>>>>>                         _______________________________________________
>>>>>>>                         wildfly-dev mailing list
>>>>>>>                         [hidden email]
>>>>>>>                         <mailto:[hidden email]>
>>>>>>>                         https://lists.jboss.org/mailman/listinfo/wildfly-dev
>>>>>>
>>>>>>
>>>>>
>>>>
>>>>
>>>
>>     _______________________________________________
>>     wildfly-dev mailing list
>>     [hidden email] <mailto:[hidden email]>
>>     https://lists.jboss.org/mailman/listinfo/wildfly-dev
>
>     --
>     Jason T. Greene
>     WildFly Lead / JBoss EAP Platform Architect
>     JBoss, a division of Red Hat
>
>
>
>
> _______________________________________________
> wildfly-dev mailing list
> [hidden email]
> https://lists.jboss.org/mailman/listinfo/wildfly-dev
>
_______________________________________________
wildfly-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/wildfly-dev
Reply | Threaded
Open this post in threaded view
|

Re: HTTP/2 out of the box in Wildfly 10.1

Darran Lofthouse
In reply to this post by Stuart Douglas
For schema changes I find it better to split out the forking of the
schema and then the configuration changes into separate commits - that
way we can actually see what changed which is essential when porting
upstream which will require changes to version 5 of the schema that
already exists.  Is there any chance we can split this into two?

Darran.

On 08/06/16 02:51, Stuart Douglas wrote:

> I have created a PR for this here:
> https://github.com/wildfly/wildfly-core/pull/1596 (it will also require
> some upstream changes).
>
> Basically this just creates a new schema version, and add the
> 'generate-self-signed-certificate-host' attribute to the keystore.
>
> I have not added a script to enable HTTPS over management as Jason
> suggested, I am not 100% sure if that really belongs in core or as part
> of the full distribution.
>
> Stuart
>
> On Wed, Jun 8, 2016 at 6:55 AM, Stuart Douglas
> <[hidden email] <mailto:[hidden email]>> wrote:
>
>
>
>     On Wed, Jun 8, 2016 at 6:51 AM, Jason Greene
>     <[hidden email] <mailto:[hidden email]>> wrote:
>
>         So after reviewing this thread and discussing with a few folks,
>         I’d like to propose, for 10.1:
>
>         #1b - Same as the previous #1, we don’t enable TLS for
>         management by default for now, but we additionally include an
>         extra cli script to enable TLS.
>
>
>     We would leave the cert generation bit in the security realm, but
>     just don't enable the HTTPS interface. That way all that is required
>     is for the user to add the https="managements-https" attribute.
>
>     Stuart
>
>
>
>         For 11 I think we should move to TLS by default, perhaps with a
>         configurable URL policy on redirects, and address the
>         incongruence with upgrade over app.
>
>         I think its likely reasonable to redirect by default for 11, but
>         we can hash that out further. One nice thing I had forgotten
>         about is that the JDK will prompt for you to add unknown certs,
>         and this all works with the CLI[1]. So it’s really only
>         non-interactive clients we have to worry about, and that sounds
>         like a reasonable burden for upgrade.
>
>         [1]
>
>         [disconnected /] connect
>         Unable to connect due to unrecognised server certificate
>         Subject    - CN=foo,OU=foo,L=Madison,ST=WI,C=US
>         Issuer     - CN=myServer, OU=test, L=Madison, ST=WI, C=US
>         Valid From - Tue Jun 07 15:22:06 CDT 2016
>         Valid To   - Thu Jun 07 15:22:06 CDT 2018
>         MD5 : cd:68:be:0b:e0:c0:1c:63:d5:2a:85:c8:d1:9d:e7:7d
>         SHA1 : ae:f8:35:fd:09:c9:b3:08:05:59:a6:40:5e:ac:6e:e8:ce:85:72:4b
>
>         Accept certificate? [N]o, [T]emporarily, [P]ermenantly : t
>
>
>
>>         On Jun 7, 2016, at 6:24 AM, Jason T. Greene
>>         <[hidden email] <mailto:[hidden email]>> wrote:
>>
>>         Long term I think we want management using TLS, but that can
>>         of course come in phases. Assuming 2) is one of those phases
>>         to come (either now or later), a following step is that the
>>         CLI, and really any remoting client, should prefer TLS with a
>>         defaulted trust store location that points to the keystore.
>>
>>         With 2) if we have the default of the attribute that forces
>>         redirect be true, and our default config be false, then
>>         someone that carries over their old config would not have a
>>         potential security weakness. If they have a CLI script that
>>         adds the https port, it will fail, hopefully sending a signal
>>         to look. Although, the user might just assume that oh it's
>>         there, I don't have to do anything.
>>
>>         Another interesting thing about 2 is that IIRC we have
>>         conflicting behavior between the app port which doesn't force
>>         upgrade and the management port which does.
>>
>>         So my preference is 2, because at some point we have to do it
>>         anyway, and if we have TLS out of the box might as well use it.
>>
>>         On Jun 6, 2016, at 10:48 PM, Stuart Douglas
>>         <[hidden email]
>>         <mailto:[hidden email]>> wrote:
>>
>>>         So while implementing this I have noticed a potential problem
>>>         that it would be good to get some feedback on.
>>>
>>>         If the management interface has SSL by default then the HTTP
>>>         interface will always redirect to the HTTPS interface. This
>>>         effectively breaks the management API, as clients such as the
>>>         CLI, Arquillian etc will be redirected to HTTPS, and then
>>>         reject the self signed certificate (as they should).
>>>
>>>         I am not sure what to do about this, these are the options as
>>>         I see them:
>>>
>>>         1) Don't enable SSL for the management interface (just for
>>>         the Undertow subsystem). The management interface can still
>>>         use this auto-generation capability, it just won't be enable
>>>         by default (we could even leave the cert in the security
>>>         domain, but just not enable the https interface).
>>>
>>>         2) Disable automatic redirects for HTTP upgrade requests
>>>         (potentially controlled by an attribute). This will allow the
>>>         CLI etc to work, but at the price of potentially reducing
>>>         security, as some connections that would have previously been
>>>         redirected to use HTTPS will no longer do this.
>>>
>>>         3) Enable it by default and leave it broken. We can setup
>>>         some kind of automatic trust store thing so the local CLI
>>>         works, and can get our test suite to work with Arquillian in
>>>         a similar manner. Personally I think this is a terrible idea,
>>>         but I am including it for completeness.
>>>
>>>         Personally I think we should go for 1). Given that this is
>>>         supposed to be about developer usability I don't think having
>>>         management also use SSL as being that important.
>>>
>>>         Stuart
>>>
>>>         On Mon, Jun 6, 2016 at 10:24 PM, Jason T. Greene
>>>         <[hidden email] <mailto:[hidden email]>> wrote:
>>>
>>>             Awesome! Another idea I had on how we could get away with
>>>             it being in server boot, is to have a pre-boot first time
>>>             setup task, either launched from the shell/batch scripts
>>>             or as a special pre-step before the AS module loads. We
>>>             could then report boot time as the time AFTER first time
>>>             installation tasks have completed, which I think is fair
>>>             because the server hasn't yet been started.
>>>
>>>             On Jun 5, 2016, at 11:53 PM, Stuart Douglas
>>>             <[hidden email]
>>>             <mailto:[hidden email]>> wrote:
>>>
>>>>             I have some initial work on this at:
>>>>             https://github.com/stuartwdouglas/wildfly-core/tree/WFCORE-1576
>>>>
>>>>             If you go to https://localhost:9993
>>>>             <https://localhost:9993/> it will generate the
>>>>             certificate (although all that will be served is a 404
>>>>             page as the console is not installed).
>>>>
>>>>             Stuart
>>>>
>>>>             On Mon, Jun 6, 2016 at 12:46 PM, Stuart Douglas
>>>>             <[hidden email]
>>>>             <mailto:[hidden email]>> wrote:
>>>>
>>>>                 I think that would actually end up being more complex.
>>>>
>>>>                 Stuart
>>>>
>>>>                 On Mon, Jun 6, 2016 at 12:45 PM, Jason T. Greene
>>>>                 <[hidden email]
>>>>                 <mailto:[hidden email]>> wrote:
>>>>
>>>>                     Another option could be a post boot task. So
>>>>                     it's still eager but don't block completed
>>>>                     start. We'd still need to block Tls ports
>>>>                     though. So maybe this does not help
>>>>
>>>>                     On Jun 5, 2016, at 9:31 PM, Stuart Douglas
>>>>                     <[hidden email]
>>>>                     <mailto:[hidden email]>> wrote:
>>>>
>>>>>                     2048 bits adds close to a second to first boot
>>>>>                     on my machine (obviously subsequent boots are
>>>>>                     unaffected).
>>>>>
>>>>>                     This is probably a bit much, I will work on
>>>>>                     getting a POC for the lazy loading approach
>>>>>                     implemented.
>>>>>
>>>>>                     Stuart
>>>>>
>>>>>                     On Mon, Jun 6, 2016 at 12:27 PM, Jason T.
>>>>>                     Greene <[hidden email]
>>>>>                     <mailto:[hidden email]>> wrote:
>>>>>
>>>>>                         We should really be generating 2048 bit keys.
>>>>>
>>>>>                         I don't like adding to our boot time, we
>>>>>                         have already seen it grow and this would be
>>>>>                         yet another case.
>>>>>
>>>>>                         On Jun 5, 2016, at 8:57 PM, Stuart Douglas
>>>>>                         <[hidden email]
>>>>>                         <mailto:[hidden email]>> wrote:
>>>>>
>>>>>>                         So I just did up a very quick prototype
>>>>>>                         that generates self signed certificates on
>>>>>>                         startup and it looks like the difference
>>>>>>                         in startup time is negligible (at least
>>>>>>                         when generating 1024 bit RSA keys). Even
>>>>>>                         if the difference is measurable it only
>>>>>>                         affects the very first startup.
>>>>>>
>>>>>>                         I think that in order to simplify the
>>>>>>                         implementation of this it may be better to
>>>>>>                         simply generate the key of first startup,
>>>>>>                         instead of attempting to do it lazily.
>>>>>>
>>>>>>                         Stuart
>>>>>>
>>>>>>                         On Sat, Jun 4, 2016 at 12:09 AM, Jason T.
>>>>>>                         Greene <[hidden email]
>>>>>>                         <mailto:[hidden email]>> wrote:
>>>>>>
>>>>>>
>>>>>>>                                 What will be default keysize? It
>>>>>>>                                 has to be probably choosen to
>>>>>>>                                 work also without "Java
>>>>>>>                                 Cryptography Extension (JCE)
>>>>>>>                                 Unlimited Strength Jurisdiction
>>>>>>>                                 Policy"
>>>>>>>
>>>>>>>
>>>>>>>                             Probably the largest that is
>>>>>>>                             supported without JCE. It does not
>>>>>>>                             matter that much, self signed certs
>>>>>>>                             are inherently insecure, this is a
>>>>>>>                             developer usability feature, not
>>>>>>>                             something that can be used in production.
>>>>>>
>>>>>>                             IIRC there is actually no limit on RSA
>>>>>>                             key size, it's only symmetric algs
>>>>>>                             that are limited, so we could use a
>>>>>>                             standard 2048 bit key without issue.
>>>>>>
>>>>>>
>>>>>>>
>>>>>>>                             Stuart
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>                                 On Thu, Jun 2, 2016 at 10:01 PM,
>>>>>>>                                 Stuart Douglas
>>>>>>>                                 <[hidden email]
>>>>>>>                                 <mailto:[hidden email]>>
>>>>>>>                                 wrote:
>>>>>>>
>>>>>>>                                     So I guess we should talk
>>>>>>>                                     about how this should
>>>>>>>                                     actually work.
>>>>>>>
>>>>>>>                                     In terms of auto generating
>>>>>>>                                     the key I was thinking we
>>>>>>>                                     would need to add a new
>>>>>>>                                     attribute to the 'keystore'
>>>>>>>                                     element under the security
>>>>>>>                                     realm, something like
>>>>>>>                                     'auto-generate-cert-host="localhost"'.
>>>>>>>                                     I am not sure what other
>>>>>>>                                     options we would need, or how
>>>>>>>                                     configurable we should make
>>>>>>>                                     it, but as this is for
>>>>>>>                                     testing/development purposes
>>>>>>>                                     I don't think we need to
>>>>>>>                                     expose full control over the
>>>>>>>                                     certificate generation process.
>>>>>>>
>>>>>>>                                     In terms of the
>>>>>>>                                     implementation we could just
>>>>>>>                                     implement an SSLContext
>>>>>>>                                     wrapper, that can do the
>>>>>>>                                     generation and then create a
>>>>>>>                                     'real' SSLContext the first
>>>>>>>                                     time it is asked to create
>>>>>>>                                     and SSLEngine.
>>>>>>>
>>>>>>>                                     Stuart
>>>>>>>
>>>>>>>                                     On Fri, Jun 3, 2016 at 3:19
>>>>>>>                                     AM, Jason Greene
>>>>>>>                                     <[hidden email]
>>>>>>>                                     <mailto:[hidden email]>>
>>>>>>>                                     wrote:
>>>>>>>
>>>>>>>
>>>>>>>                                         > On Jun 2, 2016, at 11:29 AM, Harold Campbell <[hidden email]
>>>>>>>                                         <mailto:[hidden email]>> wrote:
>>>>>>>                                         >
>>>>>>>                                         > On Thu, 2016-06-02 at 09:22 +1000, Stuart Douglas wrote:
>>>>>>>                                         >> Hi All,
>>>>>>>                                         >>
>>>>>>>                                         >> I would like to propose that we add support for HTTP/2 out of the box
>>>>>>>                                         >> in Wildfly 10.1.
>>>>>>>                                         >>
>>>>>>>                                         >
>>>>>>>                                         > This lowly user desperately wants a release containing the fix to WFLY-
>>>>>>>                                         > 6283 sooner rather than later. I'm sure other people have other pet
>>>>>>>                                         > bugs awaiting release.
>>>>>>>                                         >
>>>>>>>                                         > I have no opinion on HTTP/2 being added other than to ask that pent up
>>>>>>>                                         > bug fixes be kept in mind.
>>>>>>>
>>>>>>>
>>>>>>>                                         Hi Harold,
>>>>>>>
>>>>>>>                                         That fix is already in
>>>>>>>                                         master, so it will be
>>>>>>>                                         included in 10.1.
>>>>>>>
>>>>>>>                                         --
>>>>>>>                                         Jason T. Greene
>>>>>>>                                         WildFly Lead / JBoss EAP
>>>>>>>                                         Platform Architect
>>>>>>>                                         JBoss, a division of Red Hat
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>                                     _______________________________________________
>>>>>>>                                     wildfly-dev mailing list
>>>>>>>                                     [hidden email]
>>>>>>>                                     <mailto:[hidden email]>
>>>>>>>                                     https://lists.jboss.org/mailman/listinfo/wildfly-dev
>>>>>>>
>>>>>>>                             _______________________________________________
>>>>>>>                             wildfly-dev mailing list
>>>>>>>                             [hidden email]
>>>>>>>                             <mailto:[hidden email]>
>>>>>>>                             https://lists.jboss.org/mailman/listinfo/wildfly-dev
>>>>>>
>>>>>>
>>>>>
>>>>
>>>>
>>>
>>         _______________________________________________
>>         wildfly-dev mailing list
>>         [hidden email] <mailto:[hidden email]>
>>         https://lists.jboss.org/mailman/listinfo/wildfly-dev
>
>         --
>         Jason T. Greene
>         WildFly Lead / JBoss EAP Platform Architect
>         JBoss, a division of Red Hat
>
>
>
>
>
> _______________________________________________
> wildfly-dev mailing list
> [hidden email]
> https://lists.jboss.org/mailman/listinfo/wildfly-dev
>
_______________________________________________
wildfly-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/wildfly-dev
Reply | Threaded
Open this post in threaded view
|

Re: HTTP/2 out of the box in Wildfly 10.1

jtgreene
Administrator
In reply to this post by Darran Lofthouse



> On Jun 8, 2016, at 5:23 AM, Darran Lofthouse <[hidden email]> wrote:
>
>
>
>> On 07/06/16 21:51, Jason Greene wrote:
>> So after reviewing this thread and discussing with a few folks, I’d like
>> to propose, for 10.1:
>>
>> #1b - Same as the previous #1, we don’t enable TLS for management by
>> default for now, but we additionally include an extra cli script to
>> enable TLS.
>>
>> For 11 I think we should move to TLS by default, perhaps with a
>> configurable URL policy on redirects, and address the incongruence with
>> upgrade over app.
>>
>> I think its likely reasonable to redirect by default for 11, but we can
>> hash that out further. One nice thing I had forgotten about is that the
>> JDK will prompt for you to add unknown certs, and this all works with
>> the CLI[1]. So it’s really only non-interactive clients we have to worry
>> about, and that sounds like a reasonable burden for upgrade.
>>
>> [1]
>
> That is not the JDK, that is the code I worked on to create a more
> intuitive user experience when the CLI encounters an unexpected
> certificate ;-)


Ah duh. Well that explains it. I misread the stack trace. This is exactly what I thought we should do for 11 as part of enabling this by default.

>
>>
>> [disconnected /] connect
>> Unable to connect due to unrecognised server certificate
>> Subject    - CN=foo,OU=foo,L=Madison,ST=WI,C=US
>> Issuer     - CN=myServer, OU=test, L=Madison, ST=WI, C=US
>> Valid From - Tue Jun 07 15:22:06 CDT 2016
>> Valid To   - Thu Jun 07 15:22:06 CDT 2018
>> MD5 : cd:68:be:0b:e0:c0:1c:63:d5:2a:85:c8:d1:9d:e7:7d
>> SHA1 : ae:f8:35:fd:09:c9:b3:08:05:59:a6:40:5e:ac:6e:e8:ce:85:72:4b
>>
>> Accept certificate? [N]o, [T]emporarily, [P]ermenantly : t
>>
>>
>>> On Jun 7, 2016, at 6:24 AM, Jason T. Greene <[hidden email]
>>> <mailto:[hidden email]>> wrote:
>>>
>>> Long term I think we want management using TLS, but that can of course
>>> come in phases. Assuming 2) is one of those phases to come (either now
>>> or later), a following step is that the CLI, and really any remoting
>>> client, should prefer TLS with a defaulted trust store location that
>>> points to the keystore.
>>>
>>> With 2) if we have the default of the attribute that forces redirect
>>> be true, and our default config be false, then someone that carries
>>> over their old config would not have a potential security weakness. If
>>> they have a CLI script that adds the https port, it will fail,
>>> hopefully sending a signal to look. Although, the user might just
>>> assume that oh it's there, I don't have to do anything.
>>>
>>> Another interesting thing about 2 is that IIRC we have conflicting
>>> behavior between the app port which doesn't force upgrade and the
>>> management port which does.
>>>
>>> So my preference is 2, because at some point we have to do it anyway,
>>> and if we have TLS out of the box might as well use it.
>>>
>>> On Jun 6, 2016, at 10:48 PM, Stuart Douglas
>>> <[hidden email] <mailto:[hidden email]>> wrote:
>>>
>>>> So while implementing this I have noticed a potential problem that it
>>>> would be good to get some feedback on.
>>>>
>>>> If the management interface has SSL by default then the HTTP
>>>> interface will always redirect to the HTTPS interface. This
>>>> effectively breaks the management API, as clients such as the CLI,
>>>> Arquillian etc will be redirected to HTTPS, and then reject the self
>>>> signed certificate (as they should).
>>>>
>>>> I am not sure what to do about this, these are the options as I see them:
>>>>
>>>> 1) Don't enable SSL for the management interface (just for the
>>>> Undertow subsystem). The management interface can still use this
>>>> auto-generation capability, it just won't be enable by default (we
>>>> could even leave the cert in the security domain, but just not enable
>>>> the https interface).
>>>>
>>>> 2) Disable automatic redirects for HTTP upgrade requests (potentially
>>>> controlled by an attribute). This will allow the CLI etc to work, but
>>>> at the price of potentially reducing security, as some connections
>>>> that would have previously been redirected to use HTTPS will no
>>>> longer do this.
>>>>
>>>> 3) Enable it by default and leave it broken. We can setup some kind
>>>> of automatic trust store thing so the local CLI works, and can get
>>>> our test suite to work with Arquillian in a similar manner.
>>>> Personally I think this is a terrible idea, but I am including it for
>>>> completeness.
>>>>
>>>> Personally I think we should go for 1). Given that this is supposed
>>>> to be about developer usability I don't think having management also
>>>> use SSL as being that important.
>>>>
>>>> Stuart
>>>>
>>>> On Mon, Jun 6, 2016 at 10:24 PM, Jason T. Greene
>>>> <[hidden email] <mailto:[hidden email]>> wrote:
>>>>
>>>>    Awesome! Another idea I had on how we could get away with it
>>>>    being in server boot, is to have a pre-boot first time setup
>>>>    task, either launched from the shell/batch scripts or as a
>>>>    special pre-step before the AS module loads. We could then report
>>>>    boot time as the time AFTER first time installation tasks have
>>>>    completed, which I think is fair because the server hasn't yet
>>>>    been started.
>>>>
>>>>    On Jun 5, 2016, at 11:53 PM, Stuart Douglas
>>>>    <[hidden email] <mailto:[hidden email]>>
>>>>    wrote:
>>>>
>>>>>    I have some initial work on this at:
>>>>>    https://github.com/stuartwdouglas/wildfly-core/tree/WFCORE-1576
>>>>>
>>>>>    If you go to https://localhost:9993 <https://localhost:9993/> it
>>>>>    will generate the certificate (although all that will be served
>>>>>    is a 404 page as the console is not installed).
>>>>>
>>>>>    Stuart
>>>>>
>>>>>    On Mon, Jun 6, 2016 at 12:46 PM, Stuart Douglas
>>>>>    <[hidden email] <mailto:[hidden email]>>
>>>>>    wrote:
>>>>>
>>>>>        I think that would actually end up being more complex.
>>>>>
>>>>>        Stuart
>>>>>
>>>>>        On Mon, Jun 6, 2016 at 12:45 PM, Jason T. Greene
>>>>>        <[hidden email] <mailto:[hidden email]>>
>>>>>        wrote:
>>>>>
>>>>>            Another option could be a post boot task. So it's still
>>>>>            eager but don't block completed start. We'd still need
>>>>>            to block Tls ports though. So maybe this does not help
>>>>>
>>>>>            On Jun 5, 2016, at 9:31 PM, Stuart Douglas
>>>>>            <[hidden email]
>>>>>            <mailto:[hidden email]>> wrote:
>>>>>
>>>>>>            2048 bits adds close to a second to first boot on my
>>>>>>            machine (obviously subsequent boots are unaffected).
>>>>>>
>>>>>>            This is probably a bit much, I will work on getting a
>>>>>>            POC for the lazy loading approach implemented.
>>>>>>
>>>>>>            Stuart
>>>>>>
>>>>>>            On Mon, Jun 6, 2016 at 12:27 PM, Jason T. Greene
>>>>>>            <[hidden email]
>>>>>>            <mailto:[hidden email]>> wrote:
>>>>>>
>>>>>>                We should really be generating 2048 bit keys.
>>>>>>
>>>>>>                I don't like adding to our boot time, we have
>>>>>>                already seen it grow and this would be yet another
>>>>>>                case.
>>>>>>
>>>>>>                On Jun 5, 2016, at 8:57 PM, Stuart Douglas
>>>>>>                <[hidden email]
>>>>>>                <mailto:[hidden email]>> wrote:
>>>>>>
>>>>>>>                So I just did up a very quick prototype that
>>>>>>>                generates self signed certificates on startup and
>>>>>>>                it looks like the difference in startup time is
>>>>>>>                negligible (at least when generating 1024 bit RSA
>>>>>>>                keys). Even if the difference is measurable it
>>>>>>>                only affects the very first startup.
>>>>>>>
>>>>>>>                I think that in order to simplify the
>>>>>>>                implementation of this it may be better to simply
>>>>>>>                generate the key of first startup, instead of
>>>>>>>                attempting to do it lazily.
>>>>>>>
>>>>>>>                Stuart
>>>>>>>
>>>>>>>                On Sat, Jun 4, 2016 at 12:09 AM, Jason T. Greene
>>>>>>>                <[hidden email]
>>>>>>>                <mailto:[hidden email]>> wrote:
>>>>>>>
>>>>>>>
>>>>>>>>                        What will be default keysize? It has to
>>>>>>>>                        be probably choosen to work also without
>>>>>>>>                        "Java Cryptography Extension (JCE)
>>>>>>>>                        Unlimited Strength Jurisdiction Policy"
>>>>>>>>
>>>>>>>>
>>>>>>>>                    Probably the largest that is supported
>>>>>>>>                    without JCE. It does not matter that much,
>>>>>>>>                    self signed certs are inherently insecure,
>>>>>>>>                    this is a developer usability feature, not
>>>>>>>>                    something that can be used in production.
>>>>>>>
>>>>>>>                    IIRC there is actually no limit on RSA key
>>>>>>>                    size, it's only symmetric algs that are
>>>>>>>                    limited, so we could use a standard 2048 bit
>>>>>>>                    key without issue.
>>>>>>>
>>>>>>>
>>>>>>>>
>>>>>>>>                    Stuart
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>                        On Thu, Jun 2, 2016 at 10:01 PM, Stuart
>>>>>>>>                        Douglas <[hidden email]
>>>>>>>>                        <mailto:[hidden email]>> wrote:
>>>>>>>>
>>>>>>>>                            So I guess we should talk about how
>>>>>>>>                            this should actually work.
>>>>>>>>
>>>>>>>>                            In terms of auto generating the key I
>>>>>>>>                            was thinking we would need to add a
>>>>>>>>                            new attribute to the 'keystore'
>>>>>>>>                            element under the security realm,
>>>>>>>>                            something like
>>>>>>>>                            'auto-generate-cert-host="localhost"'. I
>>>>>>>>                            am not sure what other options we
>>>>>>>>                            would need, or how configurable we
>>>>>>>>                            should make it, but as this is for
>>>>>>>>                            testing/development purposes I don't
>>>>>>>>                            think we need to expose full control
>>>>>>>>                            over the certificate generation process.
>>>>>>>>
>>>>>>>>                            In terms of the implementation we
>>>>>>>>                            could just implement an SSLContext
>>>>>>>>                            wrapper, that can do the generation
>>>>>>>>                            and then create a 'real' SSLContext
>>>>>>>>                            the first time it is asked to create
>>>>>>>>                            and SSLEngine.
>>>>>>>>
>>>>>>>>                            Stuart
>>>>>>>>
>>>>>>>>                            On Fri, Jun 3, 2016 at 3:19 AM, Jason
>>>>>>>>                            Greene <[hidden email]
>>>>>>>>                            <mailto:[hidden email]>> wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>>>> On Jun 2, 2016, at 11:29 AM, Harold Campbell <[hidden email]
>>>>>>>>>                                <mailto:[hidden email]>> wrote:
>>>>>>>>>
>>>>>>>>>> On Thu, 2016-06-02 at 09:22 +1000, Stuart Douglas wrote:
>>>>>>>>>> Hi All,
>>>>>>>>>>
>>>>>>>>>> I would like to propose that we add support for HTTP/2 out of the box
>>>>>>>>>> in Wildfly 10.1.
>>>>>>>>>
>>>>>>>>> This lowly user desperately wants a release containing the fix to WFLY-
>>>>>>>>> 6283 sooner rather than later. I'm sure other people have other pet
>>>>>>>>> bugs awaiting release.
>>>>>>>>>
>>>>>>>>> I have no opinion on HTTP/2 being added other than to ask that pent up
>>>>>>>>> bug fixes be kept in mind.
>>>>>>>>
>>>>>>>>
>>>>>>>>                                Hi Harold,
>>>>>>>>
>>>>>>>>                                That fix is already in master, so
>>>>>>>>                                it will be included in 10.1.
>>>>>>>>
>>>>>>>>                                --
>>>>>>>>                                Jason T. Greene
>>>>>>>>                                WildFly Lead / JBoss EAP Platform
>>>>>>>>                                Architect
>>>>>>>>                                JBoss, a division of Red Hat
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>                            _______________________________________________
>>>>>>>>                            wildfly-dev mailing list
>>>>>>>>                            [hidden email]
>>>>>>>>                            <mailto:[hidden email]>
>>>>>>>>                            https://lists.jboss.org/mailman/listinfo/wildfly-dev
>>>>>>>>
>>>>>>>>                    _______________________________________________
>>>>>>>>                    wildfly-dev mailing list
>>>>>>>>                    [hidden email]
>>>>>>>>                    <mailto:[hidden email]>
>>>>>>>>                    https://lists.jboss.org/mailman/listinfo/wildfly-dev
>>> _______________________________________________
>>> wildfly-dev mailing list
>>> [hidden email] <mailto:[hidden email]>
>>> https://lists.jboss.org/mailman/listinfo/wildfly-dev
>>
>> --
>> Jason T. Greene
>> WildFly Lead / JBoss EAP Platform Architect
>> JBoss, a division of Red Hat
>>
>>
>>
>> _______________________________________________
>> wildfly-dev mailing list
>> [hidden email]
>> https://lists.jboss.org/mailman/listinfo/wildfly-dev
> _______________________________________________
> wildfly-dev mailing list
> [hidden email]
> https://lists.jboss.org/mailman/listinfo/wildfly-dev

_______________________________________________
wildfly-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/wildfly-dev
Reply | Threaded
Open this post in threaded view
|

Re: HTTP/2 out of the box in Wildfly 10.1

jtgreene
Administrator
In reply to this post by Darran Lofthouse


> On Jun 8, 2016, at 5:24 AM, Darran Lofthouse <[hidden email]> wrote:
>
>> On 07/06/16 21:55, Stuart Douglas wrote:
>>
>>
>> On Wed, Jun 8, 2016 at 6:51 AM, Jason Greene <[hidden email]
>> <mailto:[hidden email]>> wrote:
>>
>>    So after reviewing this thread and discussing with a few folks, I’d
>>    like to propose, for 10.1:
>>
>>    #1b - Same as the previous #1, we don’t enable TLS for management by
>>    default for now, but we additionally include an extra cli script to
>>    enable TLS.
>>
>>
>> We would leave the cert generation bit in the security realm, but just
>> don't enable the HTTPS interface. That way all that is required is for
>> the user to add the https="managements-https" attribute.
>
> +1 It is a minimal config change then to use it.


Indeed, but we still need to have a script so the user doesn't have to dig around for the magic setting.

>
>
>>
>> Stuart
>>
>>
>>
>>    For 11 I think we should move to TLS by default, perhaps with a
>>    configurable URL policy on redirects, and address the incongruence
>>    with upgrade over app.
>>
>>    I think its likely reasonable to redirect by default for 11, but we
>>    can hash that out further. One nice thing I had forgotten about is
>>    that the JDK will prompt for you to add unknown certs, and this all
>>    works with the CLI[1]. So it’s really only non-interactive clients
>>    we have to worry about, and that sounds like a reasonable burden for
>>    upgrade.
>>
>>    [1]
>>
>>    [disconnected /] connect
>>    Unable to connect due to unrecognised server certificate
>>    Subject    - CN=foo,OU=foo,L=Madison,ST=WI,C=US
>>    Issuer     - CN=myServer, OU=test, L=Madison, ST=WI, C=US
>>    Valid From - Tue Jun 07 15:22:06 CDT 2016
>>    Valid To   - Thu Jun 07 15:22:06 CDT 2018
>>    MD5 : cd:68:be:0b:e0:c0:1c:63:d5:2a:85:c8:d1:9d:e7:7d
>>    SHA1 : ae:f8:35:fd:09:c9:b3:08:05:59:a6:40:5e:ac:6e:e8:ce:85:72:4b
>>
>>    Accept certificate? [N]o, [T]emporarily, [P]ermenantly : t
>>
>>
>>
>>>    On Jun 7, 2016, at 6:24 AM, Jason T. Greene
>>>    <[hidden email] <mailto:[hidden email]>> wrote:
>>>
>>>    Long term I think we want management using TLS, but that can of
>>>    course come in phases. Assuming 2) is one of those phases to come
>>>    (either now or later), a following step is that the CLI, and
>>>    really any remoting client, should prefer TLS with a defaulted
>>>    trust store location that points to the keystore.
>>>
>>>    With 2) if we have the default of the attribute that forces
>>>    redirect be true, and our default config be false, then someone
>>>    that carries over their old config would not have a potential
>>>    security weakness. If they have a CLI script that adds the https
>>>    port, it will fail, hopefully sending a signal to look. Although,
>>>    the user might just assume that oh it's there, I don't have to do
>>>    anything.
>>>
>>>    Another interesting thing about 2 is that IIRC we have conflicting
>>>    behavior between the app port which doesn't force upgrade and the
>>>    management port which does.
>>>
>>>    So my preference is 2, because at some point we have to do it
>>>    anyway, and if we have TLS out of the box might as well use it.
>>>
>>>    On Jun 6, 2016, at 10:48 PM, Stuart Douglas
>>>    <[hidden email] <mailto:[hidden email]>>
>>>    wrote:
>>>
>>>>    So while implementing this I have noticed a potential problem
>>>>    that it would be good to get some feedback on.
>>>>
>>>>    If the management interface has SSL by default then the HTTP
>>>>    interface will always redirect to the HTTPS interface. This
>>>>    effectively breaks the management API, as clients such as the
>>>>    CLI, Arquillian etc will be redirected to HTTPS, and then reject
>>>>    the self signed certificate (as they should).
>>>>
>>>>    I am not sure what to do about this, these are the options as I
>>>>    see them:
>>>>
>>>>    1) Don't enable SSL for the management interface (just for the
>>>>    Undertow subsystem). The management interface can still use this
>>>>    auto-generation capability, it just won't be enable by default
>>>>    (we could even leave the cert in the security domain, but just
>>>>    not enable the https interface).
>>>>
>>>>    2) Disable automatic redirects for HTTP upgrade requests
>>>>    (potentially controlled by an attribute). This will allow the CLI
>>>>    etc to work, but at the price of potentially reducing security,
>>>>    as some connections that would have previously been redirected to
>>>>    use HTTPS will no longer do this.
>>>>
>>>>    3) Enable it by default and leave it broken. We can setup some
>>>>    kind of automatic trust store thing so the local CLI works, and
>>>>    can get our test suite to work with Arquillian in a similar
>>>>    manner. Personally I think this is a terrible idea, but I am
>>>>    including it for completeness.
>>>>
>>>>    Personally I think we should go for 1). Given that this is
>>>>    supposed to be about developer usability I don't think having
>>>>    management also use SSL as being that important.
>>>>
>>>>    Stuart
>>>>
>>>>    On Mon, Jun 6, 2016 at 10:24 PM, Jason T. Greene
>>>>    <[hidden email] <mailto:[hidden email]>> wrote:
>>>>
>>>>        Awesome! Another idea I had on how we could get away with it
>>>>        being in server boot, is to have a pre-boot first time setup
>>>>        task, either launched from the shell/batch scripts or as a
>>>>        special pre-step before the AS module loads. We could then
>>>>        report boot time as the time AFTER first time installation
>>>>        tasks have completed, which I think is fair because the
>>>>        server hasn't yet been started.
>>>>
>>>>        On Jun 5, 2016, at 11:53 PM, Stuart Douglas
>>>>        <[hidden email]
>>>>        <mailto:[hidden email]>> wrote:
>>>>
>>>>>        I have some initial work on this at:
>>>>>        https://github.com/stuartwdouglas/wildfly-core/tree/WFCORE-1576
>>>>>
>>>>>        If you go to https://localhost:9993
>>>>>        <https://localhost:9993/> it will generate the certificate
>>>>>        (although all that will be served is a 404 page as the
>>>>>        console is not installed).
>>>>>
>>>>>        Stuart
>>>>>
>>>>>        On Mon, Jun 6, 2016 at 12:46 PM, Stuart Douglas
>>>>>        <[hidden email]
>>>>>        <mailto:[hidden email]>> wrote:
>>>>>
>>>>>            I think that would actually end up being more complex.
>>>>>
>>>>>            Stuart
>>>>>
>>>>>            On Mon, Jun 6, 2016 at 12:45 PM, Jason T. Greene
>>>>>            <[hidden email]
>>>>>            <mailto:[hidden email]>> wrote:
>>>>>
>>>>>                Another option could be a post boot task. So it's
>>>>>                still eager but don't block completed start. We'd
>>>>>                still need to block Tls ports though. So maybe this
>>>>>                does not help
>>>>>
>>>>>                On Jun 5, 2016, at 9:31 PM, Stuart Douglas
>>>>>                <[hidden email]
>>>>>                <mailto:[hidden email]>> wrote:
>>>>>
>>>>>>                2048 bits adds close to a second to first boot on
>>>>>>                my machine (obviously subsequent boots are
>>>>>>                unaffected).
>>>>>>
>>>>>>                This is probably a bit much, I will work on getting
>>>>>>                a POC for the lazy loading approach implemented.
>>>>>>
>>>>>>                Stuart
>>>>>>
>>>>>>                On Mon, Jun 6, 2016 at 12:27 PM, Jason T. Greene
>>>>>>                <[hidden email]
>>>>>>                <mailto:[hidden email]>> wrote:
>>>>>>
>>>>>>                    We should really be generating 2048 bit keys.
>>>>>>
>>>>>>                    I don't like adding to our boot time, we have
>>>>>>                    already seen it grow and this would be yet
>>>>>>                    another case.
>>>>>>
>>>>>>                    On Jun 5, 2016, at 8:57 PM, Stuart Douglas
>>>>>>                    <[hidden email]
>>>>>>                    <mailto:[hidden email]>> wrote:
>>>>>>
>>>>>>>                    So I just did up a very quick prototype that
>>>>>>>                    generates self signed certificates on startup
>>>>>>>                    and it looks like the difference in startup
>>>>>>>                    time is negligible (at least when generating
>>>>>>>                    1024 bit RSA keys). Even if the difference is
>>>>>>>                    measurable it only affects the very first startup.
>>>>>>>
>>>>>>>                    I think that in order to simplify the
>>>>>>>                    implementation of this it may be better to
>>>>>>>                    simply generate the key of first startup,
>>>>>>>                    instead of attempting to do it lazily.
>>>>>>>
>>>>>>>                    Stuart
>>>>>>>
>>>>>>>                    On Sat, Jun 4, 2016 at 12:09 AM, Jason T.
>>>>>>>                    Greene <[hidden email]
>>>>>>>                    <mailto:[hidden email]>> wrote:
>>>>>>>
>>>>>>>
>>>>>>>>                            What will be default keysize? It has
>>>>>>>>                            to be probably choosen to work also
>>>>>>>>                            without "Java Cryptography Extension
>>>>>>>>                            (JCE) Unlimited Strength Jurisdiction
>>>>>>>>                            Policy"
>>>>>>>>
>>>>>>>>
>>>>>>>>                        Probably the largest that is supported
>>>>>>>>                        without JCE. It does not matter that
>>>>>>>>                        much, self signed certs are inherently
>>>>>>>>                        insecure, this is a developer usability
>>>>>>>>                        feature, not something that can be used
>>>>>>>>                        in production.
>>>>>>>
>>>>>>>                        IIRC there is actually no limit on RSA key
>>>>>>>                        size, it's only symmetric algs that are
>>>>>>>                        limited, so we could use a standard 2048
>>>>>>>                        bit key without issue.
>>>>>>>
>>>>>>>
>>>>>>>>
>>>>>>>>                        Stuart
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>                            On Thu, Jun 2, 2016 at 10:01 PM,
>>>>>>>>                            Stuart Douglas
>>>>>>>>                            <[hidden email]
>>>>>>>>                            <mailto:[hidden email]>>
>>>>>>>>                            wrote:
>>>>>>>>
>>>>>>>>                                So I guess we should talk about
>>>>>>>>                                how this should actually work.
>>>>>>>>
>>>>>>>>                                In terms of auto generating the
>>>>>>>>                                key I was thinking we would need
>>>>>>>>                                to add a new attribute to the
>>>>>>>>                                'keystore' element under the
>>>>>>>>                                security realm, something like
>>>>>>>>                                'auto-generate-cert-host="localhost"'.
>>>>>>>>                                I am not sure what other options
>>>>>>>>                                we would need, or how
>>>>>>>>                                configurable we should make it,
>>>>>>>>                                but as this is for
>>>>>>>>                                testing/development purposes I
>>>>>>>>                                don't think we need to expose
>>>>>>>>                                full control over the certificate
>>>>>>>>                                generation process.
>>>>>>>>
>>>>>>>>                                In terms of the implementation we
>>>>>>>>                                could just implement an
>>>>>>>>                                SSLContext wrapper, that can do
>>>>>>>>                                the generation and then create a
>>>>>>>>                                'real' SSLContext the first time
>>>>>>>>                                it is asked to create and SSLEngine.
>>>>>>>>
>>>>>>>>                                Stuart
>>>>>>>>
>>>>>>>>                                On Fri, Jun 3, 2016 at 3:19 AM,
>>>>>>>>                                Jason Greene
>>>>>>>>                                <[hidden email]
>>>>>>>>                                <mailto:[hidden email]>>
>>>>>>>>                                wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>>>> On Jun 2, 2016, at 11:29 AM, Harold Campbell <[hidden email]
>>>>>>>>>                                    <mailto:[hidden email]>> wrote:
>>>>>>>>>
>>>>>>>>>> On Thu, 2016-06-02 at 09:22 +1000, Stuart Douglas wrote:
>>>>>>>>>> Hi All,
>>>>>>>>>>
>>>>>>>>>> I would like to propose that we add support for HTTP/2 out of the box
>>>>>>>>>> in Wildfly 10.1.
>>>>>>>>>
>>>>>>>>> This lowly user desperately wants a release containing the fix to WFLY-
>>>>>>>>> 6283 sooner rather than later. I'm sure other people have other pet
>>>>>>>>> bugs awaiting release.
>>>>>>>>>
>>>>>>>>> I have no opinion on HTTP/2 being added other than to ask that pent up
>>>>>>>>> bug fixes be kept in mind.
>>>>>>>>
>>>>>>>>
>>>>>>>>                                    Hi Harold,
>>>>>>>>
>>>>>>>>                                    That fix is already in
>>>>>>>>                                    master, so it will be
>>>>>>>>                                    included in 10.1.
>>>>>>>>
>>>>>>>>                                    --
>>>>>>>>                                    Jason T. Greene
>>>>>>>>                                    WildFly Lead / JBoss EAP
>>>>>>>>                                    Platform Architect
>>>>>>>>                                    JBoss, a division of Red Hat
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>                                _______________________________________________
>>>>>>>>                                wildfly-dev mailing list
>>>>>>>>                                [hidden email]
>>>>>>>>                                <mailto:[hidden email]>
>>>>>>>>                                https://lists.jboss.org/mailman/listinfo/wildfly-dev
>>>>>>>>
>>>>>>>>                        _______________________________________________
>>>>>>>>                        wildfly-dev mailing list
>>>>>>>>                        [hidden email]
>>>>>>>>                        <mailto:[hidden email]>
>>>>>>>>                        https://lists.jboss.org/mailman/listinfo/wildfly-dev
>>>    _______________________________________________
>>>    wildfly-dev mailing list
>>>    [hidden email] <mailto:[hidden email]>
>>>    https://lists.jboss.org/mailman/listinfo/wildfly-dev
>>
>>    --
>>    Jason T. Greene
>>    WildFly Lead / JBoss EAP Platform Architect
>>    JBoss, a division of Red Hat
>>
>>
>>
>>
>> _______________________________________________
>> wildfly-dev mailing list
>> [hidden email]
>> https://lists.jboss.org/mailman/listinfo/wildfly-dev
> _______________________________________________
> wildfly-dev mailing list
> [hidden email]
> https://lists.jboss.org/mailman/listinfo/wildfly-dev

_______________________________________________
wildfly-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/wildfly-dev
Reply | Threaded
Open this post in threaded view
|

Re: HTTP/2 out of the box in Wildfly 10.1

jtgreene
Administrator
In reply to this post by Stuart Douglas


On Jun 7, 2016, at 8:51 PM, Stuart Douglas <[hidden email]> wrote:

I have created a PR for this here: https://github.com/wildfly/wildfly-core/pull/1596 (it will also require some upstream changes).

Basically this just creates a new schema version, and add the 'generate-self-signed-certificate-host' attribute to the keystore.

I have not added a script to enable HTTPS over management as Jason suggested, I am not 100% sure if that really belongs in core or as part of the full distribution.


It probably belongs in core so that other layered projects/products can get it as well. However for 10.1 I think it's fine if we stick it in full for now.



Stuart

On Wed, Jun 8, 2016 at 6:55 AM, Stuart Douglas <[hidden email]> wrote:


On Wed, Jun 8, 2016 at 6:51 AM, Jason Greene <[hidden email]> wrote:
So after reviewing this thread and discussing with a few folks, I’d like to propose, for 10.1:

#1b - Same as the previous #1, we don’t enable TLS for management by default for now, but we additionally include an extra cli script to enable TLS.

We would leave the cert generation bit in the security realm, but just don't enable the HTTPS interface. That way all that is required is for the user to add the https="managements-https" attribute.

Stuart
 

For 11 I think we should move to TLS by default, perhaps with a configurable URL policy on redirects, and address the incongruence with upgrade over app.

I think its likely reasonable to redirect by default for 11, but we can hash that out further. One nice thing I had forgotten about is that the JDK will prompt for you to add unknown certs, and this all works with the CLI[1]. So it’s really only non-interactive clients we have to worry about, and that sounds like a reasonable burden for upgrade.

[1]

[disconnected /] connect
Unable to connect due to unrecognised server certificate
Subject    - CN=foo,OU=foo,L=Madison,ST=WI,C=US
Issuer     - CN=myServer, OU=test, L=Madison, ST=WI, C=US
Valid From - Tue Jun 07 15:22:06 CDT 2016
Valid To   - Thu Jun 07 15:22:06 CDT 2018
MD5 : cd:68:be:0b:e0:c0:1c:63:d5:2a:85:c8:d1:9d:e7:7d
SHA1 : ae:f8:35:fd:09:c9:b3:08:05:59:a6:40:5e:ac:6e:e8:ce:85:72:4b

Accept certificate? [N]o, [T]emporarily, [P]ermenantly : t



On Jun 7, 2016, at 6:24 AM, Jason T. Greene <[hidden email]> wrote:

Long term I think we want management using TLS, but that can of course come in phases. Assuming 2) is one of those phases to come (either now or later), a following step is that the CLI, and really any remoting client, should prefer TLS with a defaulted trust store location that points to the keystore. 

With 2) if we have the default of the attribute that forces redirect be true, and our default config be false, then someone that carries over their old config would not have a potential security weakness. If they have a CLI script that adds the https port, it will fail, hopefully sending a signal to look. Although, the user might just assume that oh it's there, I don't have to do anything. 

Another interesting thing about 2 is that IIRC we have conflicting behavior between the app port which doesn't force upgrade and the management port which does.

So my preference is 2, because at some point we have to do it anyway, and if we have TLS out of the box might as well use it. 

On Jun 6, 2016, at 10:48 PM, Stuart Douglas <[hidden email]> wrote:

So while implementing this I have noticed a potential problem that it would be good to get some feedback on.

If the management interface has SSL by default then the HTTP interface will always redirect to the HTTPS interface. This effectively breaks the management API, as clients such as the CLI, Arquillian etc will be redirected to HTTPS, and then reject the self signed certificate (as they should).

I am not sure what to do about this, these are the options as I see them:

1) Don't enable SSL for the management interface (just for the Undertow subsystem). The management interface can still use this auto-generation capability, it just won't be enable by default (we could even leave the cert in the security domain, but just not enable the https interface).

2) Disable automatic redirects for HTTP upgrade requests (potentially controlled by an attribute). This will allow the CLI etc to work, but at the price of potentially reducing security, as some connections that would have previously been redirected to use HTTPS will no longer do this.

3) Enable it by default and leave it broken. We can setup some kind of automatic trust store thing so the local CLI works, and can get our test suite to work with Arquillian in a similar manner. Personally I think this is a terrible idea, but I am including it for completeness.

Personally I think we should go for 1). Given that this is supposed to be about developer usability I don't think having management also use SSL as being that important.

Stuart

On Mon, Jun 6, 2016 at 10:24 PM, Jason T. Greene <[hidden email]> wrote:
Awesome! Another idea I had on how we could get away with it being in server boot, is to have a pre-boot first time setup task, either launched from the shell/batch scripts or as a special pre-step before the AS module loads. We could then report boot time as the time AFTER first time installation tasks have completed, which I think is fair because the server hasn't yet been started.

On Jun 5, 2016, at 11:53 PM, Stuart Douglas <[hidden email]> wrote:

If you go to https://localhost:9993 it will generate the certificate (although all that will be served is a 404 page as the console is not installed).

Stuart

On Mon, Jun 6, 2016 at 12:46 PM, Stuart Douglas <[hidden email]> wrote:
I think that would actually end up being more complex.

Stuart

On Mon, Jun 6, 2016 at 12:45 PM, Jason T. Greene <[hidden email]> wrote:
Another option could be a post boot task. So it's still eager but don't block completed start. We'd still need to block Tls ports though. So maybe this does not help

On Jun 5, 2016, at 9:31 PM, Stuart Douglas <[hidden email]> wrote:

2048 bits adds close to a second to first boot on my machine (obviously subsequent boots are unaffected).

This is probably a bit much, I will work on getting a POC for the lazy loading approach implemented.

Stuart

On Mon, Jun 6, 2016 at 12:27 PM, Jason T. Greene <[hidden email]> wrote:
We should really be generating 2048 bit keys. 

I don't like adding to our boot time, we have already seen it grow and this would be yet another case.

On Jun 5, 2016, at 8:57 PM, Stuart Douglas <[hidden email]> wrote:

So I just did up a very quick prototype that generates self signed certificates on startup and it looks like the difference in startup time is negligible (at least when generating 1024 bit RSA keys). Even if the difference is measurable it only affects the very first startup.

I think that in order to simplify the implementation of this it may be better to simply generate the key of first startup, instead of attempting to do it lazily.

Stuart

On Sat, Jun 4, 2016 at 12:09 AM, Jason T. Greene <[hidden email]> wrote:

What will be default keysize? It has to be probably choosen to work also without "Java Cryptography Extension (JCE) Unlimited Strength Jurisdiction Policy"

Probably the largest that is supported without JCE. It does not matter that much, self signed certs are inherently insecure, this is a developer usability feature, not something that can be used in production.

IIRC there is actually no limit on RSA key size, it's only symmetric algs that are limited, so we could use a standard 2048 bit key without issue.



Stuart





On Thu, Jun 2, 2016 at 10:01 PM, Stuart Douglas <[hidden email]> wrote:
So I guess we should talk about how this should actually work.

In terms of auto generating the key I was thinking we would need to add a new attribute to the 'keystore' element under the security realm, something like 'auto-generate-cert-host="localhost"'. I am not sure what other options we would need, or how configurable we should make it, but as this is for testing/development purposes I don't think we need to expose full control over the certificate generation process.

In terms of the implementation we could just implement an SSLContext wrapper, that can do the generation and then create a 'real' SSLContext the first time it is asked to create and SSLEngine.

Stuart

On Fri, Jun 3, 2016 at 3:19 AM, Jason Greene <[hidden email]> wrote:

> On Jun 2, 2016, at 11:29 AM, Harold Campbell <[hidden email]> wrote:
>
> On Thu, 2016-06-02 at 09:22 +1000, Stuart Douglas wrote:
>> Hi All,
>>
>> I would like to propose that we add support for HTTP/2 out of the box
>> in Wildfly 10.1.
>>
>
> This lowly user desperately wants a release containing the fix to WFLY-
> 6283 sooner rather than later. I'm sure other people have other pet
> bugs awaiting release.
>
> I have no opinion on HTTP/2 being added other than to ask that pent up
> bug fixes be kept in mind.


Hi Harold,

That fix is already in master, so it will be included in 10.1.

--
Jason T. Greene
WildFly Lead / JBoss EAP Platform Architect
JBoss, a division of Red Hat



_______________________________________________
wildfly-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/wildfly-dev
_______________________________________________
wildfly-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/wildfly-dev





_______________________________________________
wildfly-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/wildfly-dev

--
Jason T. Greene
WildFly Lead / JBoss EAP Platform Architect
JBoss, a division of Red Hat




_______________________________________________
wildfly-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/wildfly-dev
Reply | Threaded
Open this post in threaded view
|

Re: HTTP/2 out of the box in Wildfly 10.1

Darran Lofthouse
In reply to this post by jtgreene


On 08/06/16 13:24, Jason T. Greene wrote:

>
>
>
>> On Jun 8, 2016, at 5:23 AM, Darran Lofthouse <[hidden email]> wrote:
>>
>>
>>
>>> On 07/06/16 21:51, Jason Greene wrote:
>>> So after reviewing this thread and discussing with a few folks, I’d like
>>> to propose, for 10.1:
>>>
>>> #1b - Same as the previous #1, we don’t enable TLS for management by
>>> default for now, but we additionally include an extra cli script to
>>> enable TLS.
>>>
>>> For 11 I think we should move to TLS by default, perhaps with a
>>> configurable URL policy on redirects, and address the incongruence with
>>> upgrade over app.
>>>
>>> I think its likely reasonable to redirect by default for 11, but we can
>>> hash that out further. One nice thing I had forgotten about is that the
>>> JDK will prompt for you to add unknown certs, and this all works with
>>> the CLI[1]. So it’s really only non-interactive clients we have to worry
>>> about, and that sounds like a reasonable burden for upgrade.
>>>
>>> [1]
>>
>> That is not the JDK, that is the code I worked on to create a more
>> intuitive user experience when the CLI encounters an unexpected
>> certificate ;-)
>
>
> Ah duh. Well that explains it. I misread the stack trace. This is exactly what I thought we should do for 11 as part of enabling this by default.

We could even ship the CLI with a configured server trust store and add
the generated certificate to that store - it would not stop the prompt
from working but local CLI to local server the prompt would not even be
needed.

We need to think about how the Maven plug-in will resolve Elytron client
configuration as that is typically executed from within a build but with
a common security configuration on the way it may even be possible for
that to automatically use the local server generated trust store.

>
>>
>>>
>>> [disconnected /] connect
>>> Unable to connect due to unrecognised server certificate
>>> Subject    - CN=foo,OU=foo,L=Madison,ST=WI,C=US
>>> Issuer     - CN=myServer, OU=test, L=Madison, ST=WI, C=US
>>> Valid From - Tue Jun 07 15:22:06 CDT 2016
>>> Valid To   - Thu Jun 07 15:22:06 CDT 2018
>>> MD5 : cd:68:be:0b:e0:c0:1c:63:d5:2a:85:c8:d1:9d:e7:7d
>>> SHA1 : ae:f8:35:fd:09:c9:b3:08:05:59:a6:40:5e:ac:6e:e8:ce:85:72:4b
>>>
>>> Accept certificate? [N]o, [T]emporarily, [P]ermenantly : t
>>>
>>>
>>>> On Jun 7, 2016, at 6:24 AM, Jason T. Greene <[hidden email]
>>>> <mailto:[hidden email]>> wrote:
>>>>
>>>> Long term I think we want management using TLS, but that can of course
>>>> come in phases. Assuming 2) is one of those phases to come (either now
>>>> or later), a following step is that the CLI, and really any remoting
>>>> client, should prefer TLS with a defaulted trust store location that
>>>> points to the keystore.
>>>>
>>>> With 2) if we have the default of the attribute that forces redirect
>>>> be true, and our default config be false, then someone that carries
>>>> over their old config would not have a potential security weakness. If
>>>> they have a CLI script that adds the https port, it will fail,
>>>> hopefully sending a signal to look. Although, the user might just
>>>> assume that oh it's there, I don't have to do anything.
>>>>
>>>> Another interesting thing about 2 is that IIRC we have conflicting
>>>> behavior between the app port which doesn't force upgrade and the
>>>> management port which does.
>>>>
>>>> So my preference is 2, because at some point we have to do it anyway,
>>>> and if we have TLS out of the box might as well use it.
>>>>
>>>> On Jun 6, 2016, at 10:48 PM, Stuart Douglas
>>>> <[hidden email] <mailto:[hidden email]>> wrote:
>>>>
>>>>> So while implementing this I have noticed a potential problem that it
>>>>> would be good to get some feedback on.
>>>>>
>>>>> If the management interface has SSL by default then the HTTP
>>>>> interface will always redirect to the HTTPS interface. This
>>>>> effectively breaks the management API, as clients such as the CLI,
>>>>> Arquillian etc will be redirected to HTTPS, and then reject the self
>>>>> signed certificate (as they should).
>>>>>
>>>>> I am not sure what to do about this, these are the options as I see them:
>>>>>
>>>>> 1) Don't enable SSL for the management interface (just for the
>>>>> Undertow subsystem). The management interface can still use this
>>>>> auto-generation capability, it just won't be enable by default (we
>>>>> could even leave the cert in the security domain, but just not enable
>>>>> the https interface).
>>>>>
>>>>> 2) Disable automatic redirects for HTTP upgrade requests (potentially
>>>>> controlled by an attribute). This will allow the CLI etc to work, but
>>>>> at the price of potentially reducing security, as some connections
>>>>> that would have previously been redirected to use HTTPS will no
>>>>> longer do this.
>>>>>
>>>>> 3) Enable it by default and leave it broken. We can setup some kind
>>>>> of automatic trust store thing so the local CLI works, and can get
>>>>> our test suite to work with Arquillian in a similar manner.
>>>>> Personally I think this is a terrible idea, but I am including it for
>>>>> completeness.
>>>>>
>>>>> Personally I think we should go for 1). Given that this is supposed
>>>>> to be about developer usability I don't think having management also
>>>>> use SSL as being that important.
>>>>>
>>>>> Stuart
>>>>>
>>>>> On Mon, Jun 6, 2016 at 10:24 PM, Jason T. Greene
>>>>> <[hidden email] <mailto:[hidden email]>> wrote:
>>>>>
>>>>>    Awesome! Another idea I had on how we could get away with it
>>>>>    being in server boot, is to have a pre-boot first time setup
>>>>>    task, either launched from the shell/batch scripts or as a
>>>>>    special pre-step before the AS module loads. We could then report
>>>>>    boot time as the time AFTER first time installation tasks have
>>>>>    completed, which I think is fair because the server hasn't yet
>>>>>    been started.
>>>>>
>>>>>    On Jun 5, 2016, at 11:53 PM, Stuart Douglas
>>>>>    <[hidden email] <mailto:[hidden email]>>
>>>>>    wrote:
>>>>>
>>>>>>    I have some initial work on this at:
>>>>>>    https://github.com/stuartwdouglas/wildfly-core/tree/WFCORE-1576
>>>>>>
>>>>>>    If you go to https://localhost:9993 <https://localhost:9993/> it
>>>>>>    will generate the certificate (although all that will be served
>>>>>>    is a 404 page as the console is not installed).
>>>>>>
>>>>>>    Stuart
>>>>>>
>>>>>>    On Mon, Jun 6, 2016 at 12:46 PM, Stuart Douglas
>>>>>>    <[hidden email] <mailto:[hidden email]>>
>>>>>>    wrote:
>>>>>>
>>>>>>        I think that would actually end up being more complex.
>>>>>>
>>>>>>        Stuart
>>>>>>
>>>>>>        On Mon, Jun 6, 2016 at 12:45 PM, Jason T. Greene
>>>>>>        <[hidden email] <mailto:[hidden email]>>
>>>>>>        wrote:
>>>>>>
>>>>>>            Another option could be a post boot task. So it's still
>>>>>>            eager but don't block completed start. We'd still need
>>>>>>            to block Tls ports though. So maybe this does not help
>>>>>>
>>>>>>            On Jun 5, 2016, at 9:31 PM, Stuart Douglas
>>>>>>            <[hidden email]
>>>>>>            <mailto:[hidden email]>> wrote:
>>>>>>
>>>>>>>            2048 bits adds close to a second to first boot on my
>>>>>>>            machine (obviously subsequent boots are unaffected).
>>>>>>>
>>>>>>>            This is probably a bit much, I will work on getting a
>>>>>>>            POC for the lazy loading approach implemented.
>>>>>>>
>>>>>>>            Stuart
>>>>>>>
>>>>>>>            On Mon, Jun 6, 2016 at 12:27 PM, Jason T. Greene
>>>>>>>            <[hidden email]
>>>>>>>            <mailto:[hidden email]>> wrote:
>>>>>>>
>>>>>>>                We should really be generating 2048 bit keys.
>>>>>>>
>>>>>>>                I don't like adding to our boot time, we have
>>>>>>>                already seen it grow and this would be yet another
>>>>>>>                case.
>>>>>>>
>>>>>>>                On Jun 5, 2016, at 8:57 PM, Stuart Douglas
>>>>>>>                <[hidden email]
>>>>>>>                <mailto:[hidden email]>> wrote:
>>>>>>>
>>>>>>>>                So I just did up a very quick prototype that
>>>>>>>>                generates self signed certificates on startup and
>>>>>>>>                it looks like the difference in startup time is
>>>>>>>>                negligible (at least when generating 1024 bit RSA
>>>>>>>>                keys). Even if the difference is measurable it
>>>>>>>>                only affects the very first startup.
>>>>>>>>
>>>>>>>>                I think that in order to simplify the
>>>>>>>>                implementation of this it may be better to simply
>>>>>>>>                generate the key of first startup, instead of
>>>>>>>>                attempting to do it lazily.
>>>>>>>>
>>>>>>>>                Stuart
>>>>>>>>
>>>>>>>>                On Sat, Jun 4, 2016 at 12:09 AM, Jason T. Greene
>>>>>>>>                <[hidden email]
>>>>>>>>                <mailto:[hidden email]>> wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>>>                        What will be default keysize? It has to
>>>>>>>>>                        be probably choosen to work also without
>>>>>>>>>                        "Java Cryptography Extension (JCE)
>>>>>>>>>                        Unlimited Strength Jurisdiction Policy"
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>                    Probably the largest that is supported
>>>>>>>>>                    without JCE. It does not matter that much,
>>>>>>>>>                    self signed certs are inherently insecure,
>>>>>>>>>                    this is a developer usability feature, not
>>>>>>>>>                    something that can be used in production.
>>>>>>>>
>>>>>>>>                    IIRC there is actually no limit on RSA key
>>>>>>>>                    size, it's only symmetric algs that are
>>>>>>>>                    limited, so we could use a standard 2048 bit
>>>>>>>>                    key without issue.
>>>>>>>>
>>>>>>>>
>>>>>>>>>
>>>>>>>>>                    Stuart
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>                        On Thu, Jun 2, 2016 at 10:01 PM, Stuart
>>>>>>>>>                        Douglas <[hidden email]
>>>>>>>>>                        <mailto:[hidden email]>> wrote:
>>>>>>>>>
>>>>>>>>>                            So I guess we should talk about how
>>>>>>>>>                            this should actually work.
>>>>>>>>>
>>>>>>>>>                            In terms of auto generating the key I
>>>>>>>>>                            was thinking we would need to add a
>>>>>>>>>                            new attribute to the 'keystore'
>>>>>>>>>                            element under the security realm,
>>>>>>>>>                            something like
>>>>>>>>>                            'auto-generate-cert-host="localhost"'. I
>>>>>>>>>                            am not sure what other options we
>>>>>>>>>                            would need, or how configurable we
>>>>>>>>>                            should make it, but as this is for
>>>>>>>>>                            testing/development purposes I don't
>>>>>>>>>                            think we need to expose full control
>>>>>>>>>                            over the certificate generation process.
>>>>>>>>>
>>>>>>>>>                            In terms of the implementation we
>>>>>>>>>                            could just implement an SSLContext
>>>>>>>>>                            wrapper, that can do the generation
>>>>>>>>>                            and then create a 'real' SSLContext
>>>>>>>>>                            the first time it is asked to create
>>>>>>>>>                            and SSLEngine.
>>>>>>>>>
>>>>>>>>>                            Stuart
>>>>>>>>>
>>>>>>>>>                            On Fri, Jun 3, 2016 at 3:19 AM, Jason
>>>>>>>>>                            Greene <[hidden email]
>>>>>>>>>                            <mailto:[hidden email]>> wrote:
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>>> On Jun 2, 2016, at 11:29 AM, Harold Campbell <[hidden email]
>>>>>>>>>>                                <mailto:[hidden email]>> wrote:
>>>>>>>>>>
>>>>>>>>>>> On Thu, 2016-06-02 at 09:22 +1000, Stuart Douglas wrote:
>>>>>>>>>>> Hi All,
>>>>>>>>>>>
>>>>>>>>>>> I would like to propose that we add support for HTTP/2 out of the box
>>>>>>>>>>> in Wildfly 10.1.
>>>>>>>>>>
>>>>>>>>>> This lowly user desperately wants a release containing the fix to WFLY-
>>>>>>>>>> 6283 sooner rather than later. I'm sure other people have other pet
>>>>>>>>>> bugs awaiting release.
>>>>>>>>>>
>>>>>>>>>> I have no opinion on HTTP/2 being added other than to ask that pent up
>>>>>>>>>> bug fixes be kept in mind.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>                                Hi Harold,
>>>>>>>>>
>>>>>>>>>                                That fix is already in master, so
>>>>>>>>>                                it will be included in 10.1.
>>>>>>>>>
>>>>>>>>>                                --
>>>>>>>>>                                Jason T. Greene
>>>>>>>>>                                WildFly Lead / JBoss EAP Platform
>>>>>>>>>                                Architect
>>>>>>>>>                                JBoss, a division of Red Hat
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>                            _______________________________________________
>>>>>>>>>                            wildfly-dev mailing list
>>>>>>>>>                            [hidden email]
>>>>>>>>>                            <mailto:[hidden email]>
>>>>>>>>>                            https://lists.jboss.org/mailman/listinfo/wildfly-dev
>>>>>>>>>
>>>>>>>>>                    _______________________________________________
>>>>>>>>>                    wildfly-dev mailing list
>>>>>>>>>                    [hidden email]
>>>>>>>>>                    <mailto:[hidden email]>
>>>>>>>>>                    https://lists.jboss.org/mailman/listinfo/wildfly-dev
>>>> _______________________________________________
>>>> wildfly-dev mailing list
>>>> [hidden email] <mailto:[hidden email]>
>>>> https://lists.jboss.org/mailman/listinfo/wildfly-dev
>>>
>>> --
>>> Jason T. Greene
>>> WildFly Lead / JBoss EAP Platform Architect
>>> JBoss, a division of Red Hat
>>>
>>>
>>>
>>> _______________________________________________
>>> wildfly-dev mailing list
>>> [hidden email]
>>> https://lists.jboss.org/mailman/listinfo/wildfly-dev
>> _______________________________________________
>> wildfly-dev mailing list
>> [hidden email]
>> https://lists.jboss.org/mailman/listinfo/wildfly-dev
_______________________________________________
wildfly-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/wildfly-dev
Reply | Threaded
Open this post in threaded view
|

Re: HTTP/2 out of the box in Wildfly 10.1

Stuart Douglas
In reply to this post by jtgreene
This is now in Wildfly upstream. The server will automatically generate a self signed certificate and HTTP/2 is enabled by default.

Stuart

On Wed, Jun 8, 2016 at 10:28 PM, Jason T. Greene <[hidden email]> wrote:


On Jun 7, 2016, at 8:51 PM, Stuart Douglas <[hidden email]> wrote:

I have created a PR for this here: https://github.com/wildfly/wildfly-core/pull/1596 (it will also require some upstream changes).

Basically this just creates a new schema version, and add the 'generate-self-signed-certificate-host' attribute to the keystore.

I have not added a script to enable HTTPS over management as Jason suggested, I am not 100% sure if that really belongs in core or as part of the full distribution.


It probably belongs in core so that other layered projects/products can get it as well. However for 10.1 I think it's fine if we stick it in full for now.




Stuart

On Wed, Jun 8, 2016 at 6:55 AM, Stuart Douglas <[hidden email]> wrote:


On Wed, Jun 8, 2016 at 6:51 AM, Jason Greene <[hidden email]> wrote:
So after reviewing this thread and discussing with a few folks, I’d like to propose, for 10.1:

#1b - Same as the previous #1, we don’t enable TLS for management by default for now, but we additionally include an extra cli script to enable TLS.

We would leave the cert generation bit in the security realm, but just don't enable the HTTPS interface. That way all that is required is for the user to add the https="managements-https" attribute.

Stuart
 

For 11 I think we should move to TLS by default, perhaps with a configurable URL policy on redirects, and address the incongruence with upgrade over app.

I think its likely reasonable to redirect by default for 11, but we can hash that out further. One nice thing I had forgotten about is that the JDK will prompt for you to add unknown certs, and this all works with the CLI[1]. So it’s really only non-interactive clients we have to worry about, and that sounds like a reasonable burden for upgrade.

[1]

[disconnected /] connect
Unable to connect due to unrecognised server certificate
Subject    - CN=foo,OU=foo,L=Madison,ST=WI,C=US
Issuer     - CN=myServer, OU=test, L=Madison, ST=WI, C=US
Valid From - Tue Jun 07 15:22:06 CDT 2016
Valid To   - Thu Jun 07 15:22:06 CDT 2018
MD5 : cd:68:be:0b:e0:c0:1c:63:d5:2a:85:c8:d1:9d:e7:7d
SHA1 : ae:f8:35:fd:09:c9:b3:08:05:59:a6:40:5e:ac:6e:e8:ce:85:72:4b

Accept certificate? [N]o, [T]emporarily, [P]ermenantly : t



On Jun 7, 2016, at 6:24 AM, Jason T. Greene <[hidden email]> wrote:

Long term I think we want management using TLS, but that can of course come in phases. Assuming 2) is one of those phases to come (either now or later), a following step is that the CLI, and really any remoting client, should prefer TLS with a defaulted trust store location that points to the keystore. 

With 2) if we have the default of the attribute that forces redirect be true, and our default config be false, then someone that carries over their old config would not have a potential security weakness. If they have a CLI script that adds the https port, it will fail, hopefully sending a signal to look. Although, the user might just assume that oh it's there, I don't have to do anything. 

Another interesting thing about 2 is that IIRC we have conflicting behavior between the app port which doesn't force upgrade and the management port which does.

So my preference is 2, because at some point we have to do it anyway, and if we have TLS out of the box might as well use it. 

On Jun 6, 2016, at 10:48 PM, Stuart Douglas <[hidden email]> wrote:

So while implementing this I have noticed a potential problem that it would be good to get some feedback on.

If the management interface has SSL by default then the HTTP interface will always redirect to the HTTPS interface. This effectively breaks the management API, as clients such as the CLI, Arquillian etc will be redirected to HTTPS, and then reject the self signed certificate (as they should).

I am not sure what to do about this, these are the options as I see them:

1) Don't enable SSL for the management interface (just for the Undertow subsystem). The management interface can still use this auto-generation capability, it just won't be enable by default (we could even leave the cert in the security domain, but just not enable the https interface).

2) Disable automatic redirects for HTTP upgrade requests (potentially controlled by an attribute). This will allow the CLI etc to work, but at the price of potentially reducing security, as some connections that would have previously been redirected to use HTTPS will no longer do this.

3) Enable it by default and leave it broken. We can setup some kind of automatic trust store thing so the local CLI works, and can get our test suite to work with Arquillian in a similar manner. Personally I think this is a terrible idea, but I am including it for completeness.

Personally I think we should go for 1). Given that this is supposed to be about developer usability I don't think having management also use SSL as being that important.

Stuart

On Mon, Jun 6, 2016 at 10:24 PM, Jason T. Greene <[hidden email]> wrote:
Awesome! Another idea I had on how we could get away with it being in server boot, is to have a pre-boot first time setup task, either launched from the shell/batch scripts or as a special pre-step before the AS module loads. We could then report boot time as the time AFTER first time installation tasks have completed, which I think is fair because the server hasn't yet been started.

On Jun 5, 2016, at 11:53 PM, Stuart Douglas <[hidden email]> wrote:

If you go to https://localhost:9993 it will generate the certificate (although all that will be served is a 404 page as the console is not installed).

Stuart

On Mon, Jun 6, 2016 at 12:46 PM, Stuart Douglas <[hidden email]> wrote:
I think that would actually end up being more complex.

Stuart

On Mon, Jun 6, 2016 at 12:45 PM, Jason T. Greene <[hidden email]> wrote:
Another option could be a post boot task. So it's still eager but don't block completed start. We'd still need to block Tls ports though. So maybe this does not help

On Jun 5, 2016, at 9:31 PM, Stuart Douglas <[hidden email]> wrote:

2048 bits adds close to a second to first boot on my machine (obviously subsequent boots are unaffected).

This is probably a bit much, I will work on getting a POC for the lazy loading approach implemented.

Stuart

On Mon, Jun 6, 2016 at 12:27 PM, Jason T. Greene <[hidden email]> wrote:
We should really be generating 2048 bit keys. 

I don't like adding to our boot time, we have already seen it grow and this would be yet another case.

On Jun 5, 2016, at 8:57 PM, Stuart Douglas <[hidden email]> wrote:

So I just did up a very quick prototype that generates self signed certificates on startup and it looks like the difference in startup time is negligible (at least when generating 1024 bit RSA keys). Even if the difference is measurable it only affects the very first startup.

I think that in order to simplify the implementation of this it may be better to simply generate the key of first startup, instead of attempting to do it lazily.

Stuart

On Sat, Jun 4, 2016 at 12:09 AM, Jason T. Greene <[hidden email]> wrote:

What will be default keysize? It has to be probably choosen to work also without "Java Cryptography Extension (JCE) Unlimited Strength Jurisdiction Policy"

Probably the largest that is supported without JCE. It does not matter that much, self signed certs are inherently insecure, this is a developer usability feature, not something that can be used in production.

IIRC there is actually no limit on RSA key size, it's only symmetric algs that are limited, so we could use a standard 2048 bit key without issue.



Stuart





On Thu, Jun 2, 2016 at 10:01 PM, Stuart Douglas <[hidden email]> wrote:
So I guess we should talk about how this should actually work.

In terms of auto generating the key I was thinking we would need to add a new attribute to the 'keystore' element under the security realm, something like 'auto-generate-cert-host="localhost"'. I am not sure what other options we would need, or how configurable we should make it, but as this is for testing/development purposes I don't think we need to expose full control over the certificate generation process.

In terms of the implementation we could just implement an SSLContext wrapper, that can do the generation and then create a 'real' SSLContext the first time it is asked to create and SSLEngine.

Stuart

On Fri, Jun 3, 2016 at 3:19 AM, Jason Greene <[hidden email]> wrote:

> On Jun 2, 2016, at 11:29 AM, Harold Campbell <[hidden email]> wrote:
>
> On Thu, 2016-06-02 at 09:22 +1000, Stuart Douglas wrote:
>> Hi All,
>>
>> I would like to propose that we add support for HTTP/2 out of the box
>> in Wildfly 10.1.
>>
>
> This lowly user desperately wants a release containing the fix to WFLY-
> 6283 sooner rather than later. I'm sure other people have other pet
> bugs awaiting release.
>
> I have no opinion on HTTP/2 being added other than to ask that pent up
> bug fixes be kept in mind.


Hi Harold,

That fix is already in master, so it will be included in 10.1.

--
Jason T. Greene
WildFly Lead / JBoss EAP Platform Architect
JBoss, a division of Red Hat



_______________________________________________
wildfly-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/wildfly-dev
_______________________________________________
wildfly-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/wildfly-dev





_______________________________________________
wildfly-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/wildfly-dev

--
Jason T. Greene
WildFly Lead / JBoss EAP Platform Architect
JBoss, a division of Red Hat





_______________________________________________
wildfly-dev mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/wildfly-dev
1234