I started to put together some emails for various people asking about the status of various projects we have in the pipeline, but then I realised that it might be easier to just send this query to the list and have people answer here (default to open!)
So, if you see something below that you know the status on, please reply to this post with that information. Also if you are blocked or need help, please note that too and we can possibly get some more folks involved in helping out. ;)
1. New fedora infrastructure private cloud.
msuchy was working on this. We have 3 machines setup we were doing test installs on. We also have a equalogics storage unit setup. We ran into problems with the latest openstack version, so moved back one to try and get things going.
msuchy: whats the current status? :)
2. Hyperkitty.
abompard was working on this. We have mailman01/02 production instances all setup in ansible. There was a bunch of work going on to change the database backend and ORM and such and we wanted to wait for that work to settle out. Then we were going to start migrating lists.
abompard: whats the current status?
3. Mirrormanager2
pingou was working on this. We got really far at the FAD in december. Everything is pretty much implemented I think, we just need to test a ton and roll out to production. We have various staging instances setup already.
pingou: whats the current status?
4. Bodhi2
lmacken was working on this. I setup a bodhi02.stg on rhel7/ansible, I guess we need to finish installing bodhi2 there and hooking it up in stg? It would really be nice to land this asap before the f22 cycle really starts.
lmacken: whats the current status?
5. Mediawiki update
patrick was working on this. mediawiki is moving to a new LTS release version. We were going to migrate our setup to that and try and migrate to postgresql as well.
patrick: whats the current status?
6. paste
I don't think anyone is currently working on this. We are using an older version of sticky-notes. We need to update to the newest version and migrate to postgresql. The new version bundles stuff and will be a pain to package and get our changes re-applied to.
Help wanted. ;)
7. ask
Various people were working on this. The stopforumspam plugin in stg was causing it to not load at all. Disabling that, stg now is running the newer version at least. We need to upgrade the prod version to that new version and see if there's any spam handling stuff we can do that actually works.
Help wanted.
8. trac
No one is yet working on this one, but I guess I probibly will. We need to move hosted to rhel7, but need to sort out trac for that. Likely we need to make a parallel installable 'trac012' and all our plugins for that version. Then when a new LTS version comes out we can package that as trac14 or whatever and move on up.
Packaging help wanted.
9. FMN
ralph was working on this.
We were planning on moving everything to FMN at some point. Where are we in that. Do we want to set a deadline?
10. Anytia / release-monitoring.org
pingou was working on this
Are we close to switching this on and announcing it? Whats left?
11. pkgs upgrade
bochecha and pingou was working on this.
pkgs01.stg still needs more testing? Whats the current status? Still need to fix the allow admins to login?
kevin
On Wed, Jan 7, 2015 at 9:31 PM, Kevin Fenzi kevin@scrye.com wrote:
I started to put together some emails for various people asking about the status of various projects we have in the pipeline, but then I realised that it might be easier to just send this query to the list and have people answer here (default to open!)
So, if you see something below that you know the status on, please reply to this post with that information. Also if you are blocked or need help, please note that too and we can possibly get some more folks involved in helping out. ;)
- New fedora infrastructure private cloud.
msuchy was working on this. We have 3 machines setup we were doing test installs on. We also have a equalogics storage unit setup. We ran into problems with the latest openstack version, so moved back one to try and get things going.
msuchy: whats the current status? :)
- Hyperkitty.
abompard was working on this. We have mailman01/02 production instances all setup in ansible. There was a bunch of work going on to change the database backend and ORM and such and we wanted to wait for that work to settle out. Then we were going to start migrating lists.
abompard: whats the current status?
- Mirrormanager2
pingou was working on this. We got really far at the FAD in december. Everything is pretty much implemented I think, we just need to test a ton and roll out to production. We have various staging instances setup already.
pingou: whats the current status?
- Bodhi2
lmacken was working on this. I setup a bodhi02.stg on rhel7/ansible, I guess we need to finish installing bodhi2 there and hooking it up in stg? It would really be nice to land this asap before the f22 cycle really starts.
lmacken: whats the current status?
- Mediawiki update
patrick was working on this. mediawiki is moving to a new LTS release version. We were going to migrate our setup to that and try and migrate to postgresql as well.
patrick: whats the current status?
- paste
I don't think anyone is currently working on this. We are using an older version of sticky-notes. We need to update to the newest version and migrate to postgresql. The new version bundles stuff and will be a pain to package and get our changes re-applied to.
Help wanted. ;)
I would like to help. May be we can discuss more on this before the meeting tomorrow on #fedora-admin. I guess first step would be to package the sticky-notes to the latest version.
- ask
Various people were working on this. The stopforumspam plugin in stg was causing it to not load at all. Disabling that, stg now is running the newer version at least. We need to upgrade the prod version to that new version and see if there's any spam handling stuff we can do that actually works.
Help wanted.
- trac
No one is yet working on this one, but I guess I probibly will. We need to move hosted to rhel7, but need to sort out trac for that. Likely we need to make a parallel installable 'trac012' and all our plugins for that version. Then when a new LTS version comes out we can package that as trac14 or whatever and move on up.
Packaging help wanted.
- FMN
ralph was working on this.
We were planning on moving everything to FMN at some point. Where are we in that. Do we want to set a deadline?
- Anytia / release-monitoring.org
pingou was working on this
Are we close to switching this on and announcing it? Whats left?
- pkgs upgrade
bochecha and pingou was working on this.
pkgs01.stg still needs more testing? Whats the current status? Still need to fix the allow admins to login?
kevin
infrastructure mailing list infrastructure@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/infrastructure
Hey,
I would love to work on paste and ask. Ping me on IRC, and we can get things sorted :)
On Wed, Jan 7, 2015, 12:19 PM Aditya Patawari adimania@gmail.com wrote:
On Wed, Jan 7, 2015 at 9:31 PM, Kevin Fenzi kevin@scrye.com wrote:
I started to put together some emails for various people asking about the status of various projects we have in the pipeline, but then I realised that it might be easier to just send this query to the list and have people answer here (default to open!)
So, if you see something below that you know the status on, please reply to this post with that information. Also if you are blocked or need help, please note that too and we can possibly get some more folks involved in helping out. ;)
- New fedora infrastructure private cloud.
msuchy was working on this. We have 3 machines setup we were doing test installs on. We also have a equalogics storage unit setup. We ran into problems with the latest openstack version, so moved back one to try and get things going.
msuchy: whats the current status? :)
- Hyperkitty.
abompard was working on this. We have mailman01/02 production instances all setup in ansible. There was a bunch of work going on to change the database backend and ORM and such and we wanted to wait for that work to settle out. Then we were going to start migrating lists.
abompard: whats the current status?
- Mirrormanager2
pingou was working on this. We got really far at the FAD in december. Everything is pretty much implemented I think, we just need to test a ton and roll out to production. We have various staging instances setup already.
pingou: whats the current status?
- Bodhi2
lmacken was working on this. I setup a bodhi02.stg on rhel7/ansible, I guess we need to finish installing bodhi2 there and hooking it up in stg? It would really be nice to land this asap before the f22 cycle really starts.
lmacken: whats the current status?
- Mediawiki update
patrick was working on this. mediawiki is moving to a new LTS release version. We were going to migrate our setup to that and try and migrate to postgresql as well.
patrick: whats the current status?
- paste
I don't think anyone is currently working on this. We are using an older version of sticky-notes. We need to update to the newest version and migrate to postgresql. The new version bundles stuff and will be a pain to package and get our changes re-applied to.
Help wanted. ;)
I would like to help. May be we can discuss more on this before the meeting tomorrow on #fedora-admin. I guess first step would be to package the sticky-notes to the latest version.
- ask
Various people were working on this. The stopforumspam plugin in stg was causing it to not load at all. Disabling that, stg now is running the newer version at least. We need to upgrade the prod version to that new version and see if there's any spam handling stuff we can do that actually works.
Help wanted.
- trac
No one is yet working on this one, but I guess I probibly will. We need to move hosted to rhel7, but need to sort out trac for that. Likely we need to make a parallel installable 'trac012' and all our plugins for that version. Then when a new LTS version comes out we can package that as trac14 or whatever and move on up.
Packaging help wanted.
- FMN
ralph was working on this.
We were planning on moving everything to FMN at some point. Where are we in that. Do we want to set a deadline?
- Anytia / release-monitoring.org
pingou was working on this
Are we close to switching this on and announcing it? Whats left?
- pkgs upgrade
bochecha and pingou was working on this.
pkgs01.stg still needs more testing? Whats the current status? Still need to fix the allow admins to login?
kevin
infrastructure mailing list infrastructure@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/infrastructure
infrastructure mailing list infrastructure@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/infrastructure
On Wed, Jan 07, 2015 at 09:01:52AM -0700, Kevin Fenzi wrote:
- FMN
ralph was working on this.
We were planning on moving everything to FMN at some point. Where are we in that. Do we want to set a deadline?
I'm getting a new release ready now with lots of bugfixes, new rules you can build filters from, and a much better set of defaults. I thought I'd have it out today but got side-tracked on one last optimization. It should be out next week.
Once that release is in place we should use it and test it for a few days to make sure everything is working as expected.
Next step after that is to carry out the steps to move the packager group from opt-in to opt-out (i.e., we need to automatically subscribe all packagers and then send out an announcement).
Later, after that, we can start turning off the built-in notifications from other services. pkgdb2 (iirc) does its own notifications, but it has a config switch to disable that. bodhi2 (iirc) is expecting fmn to be ready -- it doesn't have its own notifications.
- Anytia / release-monitoring.org
pingou was working on this
Are we close to switching this on and announcing it? Whats left?
Release-monitoring.org is chugging along and doing its thing. The component that files bugs on Fedora packages when there are upstream releases is the only thing left that needs some work. It is running happily in staging. We thought it was "done" until we were discussing the cutover in #fedora-apps this week and discovered that there's one relatively major feature that needs needs to be implemented: Namely, mass-subscription of package groups.
In the existing Upstream_Release_Monitoring system (cnucnu), packages that match certain patterns like ghc-*, drupal-*, rubygem-*, etc.. are all automatically monitored. To get that same functionality in the new system, we need to take care of these three things:
- Add some code to the new release monitoring backend (dubbed 'the-new-hotness') that listens for when new packages are added to the pkgdb and in response adds their upstream pointer to release-monitoring.org with the appropriate regexes, if they're not already there. If they are already there, then they need to be "mapped" to their new downstream Fedora package. - release-monitoring.org, while functional, doesn't have quite the API to handle that.. at least, there might still be some implementing to do. - (lastly, pkgdb2 should default new packages' monitor flag to True).
I plan to work on these, but I'm keeping it on the back burner until the FMN stuff mentioned above is all done. I expect them both to be ready and in place before the F22 Alpha freeze (Feb. 24th). (As always, if anyone wants to jump on these before I get to them, please do. Better to have it done sooner.)
Il giorno mer, 07/01/2015 alle 09.01 -0700, Kevin Fenzi ha scritto:
- paste
I don't think anyone is currently working on this. We are using an older version of sticky-notes. We need to update to the newest version and migrate to postgresql. The new version bundles stuff and will be a pain to package and get our changes re-applied to.
Help wanted. ;)
I can help with this one, since I already have experience with deployment of PHP applications :).
On Sun, Jan 11, 2015 at 5:02 PM, Fabio Alessandro Locati fabiolocati@gmail.com wrote:
Il giorno mer, 07/01/2015 alle 09.01 -0700, Kevin Fenzi ha scritto:
- paste
I don't think anyone is currently working on this. We are using an older version of sticky-notes. We need to update to the newest version and migrate to postgresql. The new version bundles stuff and will be a pain to package and get our changes re-applied to.
Help wanted. ;)
I can help with this one, since I already have experience with deployment of PHP applications :).
Hi Fabio and Chaoyi,
After looking at the requirements of paste, we are good from php-symphony side and all the dependencies have already been packaged. Laravel is something which we have to look at for packaging, along with 3-4 other small php libraries. Those with experience in php packaging can have a look at [1] and pick up unpackaged libraries.
[1]: https://github.com/sayakb/sticky-notes/tree/master/vendor
Hi,
<...>
- paste
I don't think anyone is currently working on this. We are using an older version of sticky-notes. We need to update to the newest version and migrate to postgresql. The new version bundles stuff and will be a pain to package and get our
<...>
sticky-notes 1.x branch has been rewritten from scratch, and introduced new dependencies which are in [1], if we could package those dependencies, I'll be able to update the package to 1.x .
Note that we already have some symfony packages but most of them are orphaned [2]
Update from 0.4.x is supported with mysql, although in our case it's a new deployment with pgsql (I guess we keep the old paste.f.o read-only for a while and then switch).
[1] https://github.com/sayakb/sticky-notes/tree/master/vendor
[2] https://admin.fedoraproject.org/pkgdb/packages/*symfony*/
Thanks in advance,
Regards.
-- Athmane
On Wed, Jan 07, 2015 at 09:01:52AM -0700, Kevin Fenzi wrote:
- Mirrormanager2
pingou was working on this. We got really far at the FAD in december. Everything is pretty much implemented I think, we just need to test a ton and roll out to production. We have various staging instances setup already.
pingou: whats the current status?
Sorry for the late reply. MirrorManager2 is in a decent state I believe. I have been working on unit-tests for its UI, fixing some bugs in the process. The last thing which I would like to manage is having some tests for the different cron jobs. This would make it easier to check their behaviors and debug in case of problems (such as the weird release number we currently have).
I added the proxy and varnish configuration for MirrorManager2 stg on ansible this morning, fixing access to the web app but looks like I broke apache in the process. Otherwise, mirrorlist seemed to work and we get output of some of the cron jobs so that seems to work but the more testing/validation the better.
- Anytia / release-monitoring.org
pingou was working on this
Are we close to switching this on and announcing it? Whats left?
Things are getting in order, but there are still some pieces missing as Ralph pointed out. One left for anitya itself is being able to handle the following folder structure: /2.4/ /2.4/2.4.1/ /2.4/2.4.2/ /2.5/ /2.5/2.5.1/ ... cnucnu was able to work with it, Anitya isn't at the moment and some projects are using it so it would be nice to add it back.
- pkgs upgrade
bochecha and pingou was working on this.
pkgs01.stg still needs more testing? Whats the current status? Still need to fix the allow admins to login?
So gitolite is still not completely working, one of the issue being the creation of the monthly logs: FATAL: errors found but logfile could not be created FATAL: /root/.gitolite/logs/gitolite-2015-01.log: No such file or directory FATAL: die chdir /root/.gitolite failed: No such file or directory<<newline>>
After that, we also haven't figured out how to give a shell access to admins and of course, the more testing the better :)
Note: the patch to mkbranch has been merged, so we will be able to have the git adjusted simply based on fedmsg message from pkgdb (similarly to what we have in place for shell account based on FAS messages).
Pierre
On Fri, Jan 16, 2015 at 02:29:31PM +0100, Pierre-Yves Chibon wrote:
- pkgs upgrade
bochecha and pingou was working on this.
pkgs01.stg still needs more testing? Whats the current status? Still need to fix the allow admins to login?
So gitolite is still not completely working, one of the issue being the creation of the monthly logs: FATAL: errors found but logfile could not be created FATAL: /root/.gitolite/logs/gitolite-2015-01.log: No such file or directory FATAL: die chdir /root/.gitolite failed: No such file or directory<<newline>>
Ok I have been working on this error which was coming from our genacls.sh script/cron. I turned out to be related to the user calling the command on the home of the correct user but revealed a more annoying problem: running `gitolite compile` tries to re-create the symbolic links to repo.git/hooks/update. This fails for us as we have our own update hook in place allowing us to run chained-hooks. So for the moment, I have commented out the two lines of code being annoying and it works fine but we would need a more permanent solution for this.
After that, we also haven't figured out how to give a shell access to admins and
This remains still
of course, the more testing the better :)
And this as well :)
Pierre
On Thu, Jan 22, 2015 at 12:39:01PM +0100, Pierre-Yves Chibon wrote:
On Fri, Jan 16, 2015 at 02:29:31PM +0100, Pierre-Yves Chibon wrote:
- pkgs upgrade
bochecha and pingou was working on this.
pkgs01.stg still needs more testing? Whats the current status? Still need to fix the allow admins to login?
So gitolite is still not completely working, one of the issue being the creation of the monthly logs: FATAL: errors found but logfile could not be created FATAL: /root/.gitolite/logs/gitolite-2015-01.log: No such file or directory FATAL: die chdir /root/.gitolite failed: No such file or directory<<newline>>
Ok I have been working on this error which was coming from our genacls.sh script/cron. I turned out to be related to the user calling the command on the home of the correct user but revealed a more annoying problem: running `gitolite compile` tries to re-create the symbolic links to repo.git/hooks/update. This fails for us as we have our own update hook in place allowing us to run chained-hooks. So for the moment, I have commented out the two lines of code being annoying and it works fine but we would need a more permanent solution for this.
After some work and re-work and further investigation, bochecha and I managed to fix this this morning and we got the fix validated by gitolite's upstream.
After that, we also haven't figured out how to give a shell access to admins and
This remains still
This was fixed yesterday with the help from gitolite's upstream (many thanks to him!)
of course, the more testing the better :)
And this as well :)
Since it seems to us that all is now fixed and ready, we are re-building the host from scratch and then all that is left is: testing :)
Pierre
On Fri, 2015-01-23 at 14:10 +0100, Pierre-Yves Chibon wrote:
Since it seems to us that all is now fixed and ready, we are re-building the host from scratch and then all that is left is: testing :)
So Pierre-Yves finished rebuilding the host and syncing some git data on it from prod.
And things just work. :)
So far, I've tested:
* shell access for admins (works for Pierre-Yves from sysadmin-main, works for me from sysadmin-noc)
* fedpkg clone/push, verifying that push fails for packages I don't have acls on
* git push of branches starting with "origin/", which is supposed to fail (https://fedorahosted.org/rel-eng/ticket/4071)
Still needs to be tested:
* cgit seems to not see any package
* fedpkg sources / new-sources fail (looking into this right now)
On Mon, Jan 26, 2015 at 04:12:31PM +0100, Mathieu Bridon wrote:
On Fri, 2015-01-23 at 14:10 +0100, Pierre-Yves Chibon wrote:
Since it seems to us that all is now fixed and ready, we are re-building the host from scratch and then all that is left is: testing :)
So Pierre-Yves finished rebuilding the host and syncing some git data on it from prod.
And things just work. :)
So far, I've tested:
shell access for admins (works for Pierre-Yves from sysadmin-main, works for me from sysadmin-noc)
fedpkg clone/push, verifying that push fails for packages I don't have acls on
git push of branches starting with "origin/", which is supposed to fail (https://fedorahosted.org/rel-eng/ticket/4071)
Still needs to be tested:
cgit seems to not see any package
fedpkg sources / new-sources fail (looking into this right now)
After some more fighting: is working: - shell access for admins - fedpkg clone, pull, push - Fails on package on which user does not have the ACLs - Fails on branches not allowed - Fails on branches named origin/... - cgit: http://pkgs.stg.fedoraproject.org/cgit/ - fedpkg new-sources / sources
All this with SELinux enabled.
Fails: - fedmsg-genacls.sh - fedmsg messages sent after an upload Of the two, the last one at least is still SELinux related, no clue for the first one.
So if we want to move it along, we can run w/o SELinux for the moment (as we do currently) or wait some more to debug/fix things.
Thanks to Mathieu and tfirg on #selinux for the great help on this work!
Pierre
On Wed, 28 Jan 2015 18:01:59 +0100 Pierre-Yves Chibon pingou@pingoured.fr wrote:
After some more fighting: is working:
- shell access for admins
- fedpkg clone, pull, push
- Fails on package on which user does not have the ACLs
- Fails on branches not allowed
- Fails on branches named origin/...
- cgit: http://pkgs.stg.fedoraproject.org/cgit/
- fedpkg new-sources / sources
All this with SELinux enabled.
Hurray!
Fails:
- fedmsg-genacls.sh
- fedmsg messages sent after an upload
Of the two, the last one at least is still SELinux related, no clue for the first one.
So if we want to move it along, we can run w/o SELinux for the moment (as we do currently) or wait some more to debug/fix things.
I'd really like to have selinux enabled on it.
If we can't figure it out, perhaps we could ask Dan Walsh to take a look?
Thanks to Mathieu and tfirg on #selinux for the great help on this work!
Yes, thanks for all your work on this... exciting to get nearer to deploying.
kevin
On Wed, Jan 28, 2015 at 06:01:59PM +0100, Pierre-Yves Chibon wrote:
On Mon, Jan 26, 2015 at 04:12:31PM +0100, Mathieu Bridon wrote:
On Fri, 2015-01-23 at 14:10 +0100, Pierre-Yves Chibon wrote:
Since it seems to us that all is now fixed and ready, we are re-building the host from scratch and then all that is left is: testing :)
So Pierre-Yves finished rebuilding the host and syncing some git data on it from prod.
And things just work. :)
So far, I've tested:
shell access for admins (works for Pierre-Yves from sysadmin-main, works for me from sysadmin-noc)
fedpkg clone/push, verifying that push fails for packages I don't have acls on
git push of branches starting with "origin/", which is supposed to fail (https://fedorahosted.org/rel-eng/ticket/4071)
Still needs to be tested:
cgit seems to not see any package
fedpkg sources / new-sources fail (looking into this right now)
After some more fighting: is working:
- shell access for admins
- fedpkg clone, pull, push
- Fails on package on which user does not have the ACLs
- Fails on branches not allowed
- Fails on branches named origin/...
- cgit: http://pkgs.stg.fedoraproject.org/cgit/
- fedpkg new-sources / sources
All this with SELinux enabled.
Fails:
- fedmsg-genacls.sh
This is now fixed. It was basically two permission issues, one for running genacls.sh which now needs to be run as root as it as to chown and chmod some files and the second was adjust the permissions to allow fedmsg to sudo as root to run genacls.sh
- fedmsg messages sent after an upload
Of the two, the last one at least is still SELinux related, no clue for the first one.
Remains this one :)
Pierre
On Thu, Jan 29, 2015 at 11:46:03AM +0100, Pierre-Yves Chibon wrote:
On Wed, Jan 28, 2015 at 06:01:59PM +0100, Pierre-Yves Chibon wrote:
On Mon, Jan 26, 2015 at 04:12:31PM +0100, Mathieu Bridon wrote:
On Fri, 2015-01-23 at 14:10 +0100, Pierre-Yves Chibon wrote:
Since it seems to us that all is now fixed and ready, we are re-building the host from scratch and then all that is left is: testing :)
So Pierre-Yves finished rebuilding the host and syncing some git data on it from prod.
And things just work. :)
So far, I've tested:
shell access for admins (works for Pierre-Yves from sysadmin-main, works for me from sysadmin-noc)
fedpkg clone/push, verifying that push fails for packages I don't have acls on
git push of branches starting with "origin/", which is supposed to fail (https://fedorahosted.org/rel-eng/ticket/4071)
Still needs to be tested:
cgit seems to not see any package
fedpkg sources / new-sources fail (looking into this right now)
After some more fighting: is working:
- shell access for admins
- fedpkg clone, pull, push
- Fails on package on which user does not have the ACLs
- Fails on branches not allowed
- Fails on branches named origin/...
- cgit: http://pkgs.stg.fedoraproject.org/cgit/
- fedpkg new-sources / sources
All this with SELinux enabled.
Fails:
- fedmsg-genacls.sh
This is now fixed. It was basically two permission issues, one for running genacls.sh which now needs to be run as root as it as to chown and chmod some files and the second was adjust the permissions to allow fedmsg to sudo as root to run genacls.sh
Note: this is likely related to the fact that we pretty much dropped using the gen-acls user.
Pierre
On Thu, Jan 29, 2015 at 11:46:03AM +0100, Pierre-Yves Chibon wrote:
On Wed, Jan 28, 2015 at 06:01:59PM +0100, Pierre-Yves Chibon wrote:
On Mon, Jan 26, 2015 at 04:12:31PM +0100, Mathieu Bridon wrote:
On Fri, 2015-01-23 at 14:10 +0100, Pierre-Yves Chibon wrote:
Since it seems to us that all is now fixed and ready, we are re-building the host from scratch and then all that is left is: testing :)
So Pierre-Yves finished rebuilding the host and syncing some git data on it from prod.
And things just work. :)
So far, I've tested:
shell access for admins (works for Pierre-Yves from sysadmin-main, works for me from sysadmin-noc)
fedpkg clone/push, verifying that push fails for packages I don't have acls on
git push of branches starting with "origin/", which is supposed to fail (https://fedorahosted.org/rel-eng/ticket/4071)
Still needs to be tested:
cgit seems to not see any package
fedpkg sources / new-sources fail (looking into this right now)
After some more fighting: is working:
- shell access for admins
- fedpkg clone, pull, push
- Fails on package on which user does not have the ACLs
- Fails on branches not allowed
- Fails on branches named origin/...
- cgit: http://pkgs.stg.fedoraproject.org/cgit/
- fedpkg new-sources / sources
All this with SELinux enabled.
Fails:
- fedmsg-genacls.sh
This is now fixed. It was basically two permission issues, one for running genacls.sh which now needs to be run as root as it as to chown and chmod some files and the second was adjust the permissions to allow fedmsg to sudo as root to run genacls.sh
- fedmsg messages sent after an upload
Of the two, the last one at least is still SELinux related, no clue for the first one.
Remains this one :)
And with one last SELinux boolean tunning, this is working \ó/
pkgs01.stg has been rebuilt (again) and all seems to work fine.
So if someone wants to review our change, I think we're good :)
Pierre
infrastructure@lists.fedoraproject.org
-
Aditya Patawari -
Athmane Madjoudj -
Chaoyi Zha -
Fabio Alessandro Locati -
Kevin Fenzi -
Mathieu Bridon -
Pierre-Yves Chibon -
Ralph Bean