#fedora-meeting: Infrastructure (2014-10-02)
Meeting started by nirik at 18:00:00 UTC. The full logs are available at
* aloha (nirik, 18:00:01)
* New folks introductions and Apprentice tasks. (nirik, 18:02:01)
* Applications status / discussion (nirik, 18:03:38)
* pkgdb2 update in prod today, another minor one tomorrow. (nirik,
* anitya almost ready for production. needs dns to finish getting
setup. (nirik, 18:07:52)
* taskotron almost ready for production. (nirik, 18:08:03)
* hyperkitty is working on sqlalchemy changes. Will deploy after those
land. (nirik, 18:15:47)
* bodhi2 is in jenkins now and running tests on commits (nirik,
* ACTION: pingou to coordinate with tyll on cnucnu bugzilla filing
under the new anitya setup. (nirik, 18:24:40)
* fedora mobile getting setup to be updated from the google play alpha
track. (nirik, 18:25:05)
* Sysadmin status / discussion (nirik, 18:26:27)
* mass reboot earlier this week. Most things are rebooted/updated
* kernel01/02 now have a bunch more memory. Thanks smooge for getting
that installed. (nirik, 18:30:15)
* db04 (rhel6) migrated to db-koji01 (rhel7) (nirik, 18:30:33)
* 0 guests left at telia01 (nirik, 18:30:43)
* 78 hosts left in puppet (nirik, 18:33:28)
* nagios/alerts recap (nirik, 18:33:44)
* Upcoming Tasks/Items (nirik, 18:40:02)
* LINK: https://apps.fedoraproject.org/calendar/list/infrastructure/
* Open Floor (nirik, 18:41:58)
Meeting ended at 18:47:43 UTC.
* pingou to coordinate with tyll on cnucnu bugzilla filing under the new
Action Items, by person
* pingou to coordinate with tyll on cnucnu bugzilla filing under the
new anitya setup.
People Present (lines said)
* nirik (109)
* pingou (36)
* relrod (15)
* smooge (11)
* abompard (6)
* lmacken (6)
* oddshocks (6)
* tflink (6)
* zodbot (5)
* threebean (2)
* RogerBTX (1)
* Neldogz (1)
* danielbruno (1)
* lanica (1)
* mpduty (1)
* dgilmore (1)
* puiterwijk (0)
* abadger1999 (0)
* mdomsch (0)
18:00:00 <nirik> #startmeeting Infrastructure (2014-10-02)
18:00:00 <zodbot> Meeting started Thu Oct 2 18:00:00 2014 UTC. The chair is nirik.
Information about MeetBot at http://wiki.debian.org/MeetBot
18:00:00 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
18:00:01 <nirik> #meetingname infrastructure
18:00:01 <nirik> #topic aloha
18:00:01 <nirik> #chair smooge relrod nirik abadger1999 lmacken dgilmore mdomsch
threebean pingou puiterwijk
18:00:01 <zodbot> The meeting name has been set to 'infrastructure'
18:00:01 <zodbot> Current chairs: abadger1999 dgilmore lmacken mdomsch nirik pingou
puiterwijk relrod smooge threebean
18:00:14 * threebean is here
18:00:14 * pingou
18:00:18 * tflink is here
18:00:19 * relrod here
18:00:26 * lmacken
18:00:28 * danielbruno is here
18:00:46 * oddshocks here
18:01:22 * RogerBTX present @ laundromat
18:01:33 * mpduty is here
18:01:58 <Neldogz> neldogz is here
18:02:01 <nirik> #topic New folks introductions and Apprentice tasks.
18:02:13 <nirik> any new folks like to introduce themselves today?
18:02:20 <nirik> or apprentices with questions, comments or ideas?
18:03:24 <nirik> ok. ;) moving along then...
18:03:38 <nirik> #topic Applications status / discussion
18:03:45 <nirik> any application news this week? :)
18:03:52 <pingou> new pkgdb in prod today
18:04:04 <nirik> nice. any list of changes?
18:04:11 <smooge> mgoodmday
18:04:23 * nirik wonders if we shouldn't try and mail at least infra list on updates?
just a short 'pkgdb2 updated, here's changes'
18:04:41 <pingou> I added the update_package_info script that was written a while
18:05:01 <pingou> whose goal is to update the package's info (summary and
description) from yum's info (thus the rpm)
18:05:20 <pingou> also, now the user orphaning a package will loose his/her commit
and approveacl ACL
18:05:34 <nirik> ok. I guess if they need it back they can unorphan. ;)
18:05:40 <pingou> yes :)
18:05:48 <pingou> the rest is mostly cosmetic changes in the UI
18:06:07 <pingou> there will be another release tomorrow to fix something in the
emails pkgdb send when it has a problem
18:06:16 <pingou> and I have a new fedocal release coming after that
18:06:23 <nirik> busy busy. ;)
18:06:32 <pingou> (w/ for example the possibility to send a reminder to multiple
18:06:40 <tflink> taskotron production is making progress
18:06:52 <pingou> and anitya is ready in prod, just waiting for the DNS to sync in:
18:07:06 <pingou> but I'll call for tester before calling it 1.0 :)
18:07:13 <tflink> waiting on two bits of code, learning the hard way about bits that
need to be monitored after restart etc.
18:07:18 <nirik> #info pkgdb2 update in prod today, another minor one tomorrow.
18:07:45 * dgilmore is kinda here but really not
18:07:52 <nirik> #info anitya almost ready for production. needs dns to finish
18:08:03 <nirik> #info taskotron almost ready for production.
18:08:28 <nirik> tflink: aside from those issues everything looking ok from a infra
standpoint for you? I know we need monitoring still.
18:08:54 <tflink> looking into some vpn issues ATM
18:09:04 * pingou needs to add a nagios check for anitya
18:09:10 <tflink> monitoring, backup (ticket hasn't been filed yet)
18:09:28 <nirik> ah yeah I am already backing up the db...
18:09:34 <nirik> but there's possibly more.
18:09:51 <tflink> there are some files on taskotron01.qa that will need to be backed
18:10:00 <nirik> ok.
18:10:07 * lanica is here for the infra meeting (sorry I'm late!)
18:10:14 <nirik> welcome lanica
18:10:57 <nirik> ok, anything else pending on the applications side?
18:11:29 <threebean> eh, I've been doing lots of monitoring stuff this week and
will be pushing out some other changes for anitya later.
18:11:41 <nirik> cool.
18:12:18 <oddshocks> general improvements to fedimg and ostree/atomic stuff. nothing
18:12:23 <oddshocks> making progress
18:12:37 <pingou> cool
18:12:46 <nirik> abompard: whats the current word on hyperkitty (if you happen to be
around). I'd be fine moving some lists if there's nothing we are waiting on.
18:12:52 <nirik> oddshocks: cool.
18:13:19 <abompard> nirik: actually there's currently lots going on around
Mailman's SQLAlchemy port
18:13:26 <abompard> it may be happenning soon
18:13:42 <lmacken> I got bodhi2 test suite back up and running on jenkins this week.
Lots of masher hacking as well, but that hasn't hit infra yet.
18:13:54 <abompard> so I'm focusing on that, the windows of availability for
barry to review code are small
18:13:56 <nirik> abompard: nice. :) do we want to wait for that?
18:14:15 <abompard> nirik: yes, if it happens as soon as I think it can
18:14:25 <nirik> lmacken: nice.
18:14:25 <oddshocks> abompard: as in warsaw?
18:14:30 <abompard> oddshocks: yep
18:14:33 <oddshocks> abompard: niceee
18:14:48 <nirik> lmacken: no weird disappearances on jenkins lately? did we ever
figure out what was happening with that?
18:15:03 <pingou> abompard: champagne \ó/
18:15:14 <abompard> pingou: hell yeah
18:15:33 <nirik> cool.
18:15:38 <lmacken> nirik: as far as plugins disappearing, I think the ansible
playbook overwrites them all, and we don't have the fedmsg plugin in git
18:15:47 <nirik> #info hyperkitty is working on sqlalchemy changes. Will deploy
after those land.
18:15:54 <nirik> lmacken: ah, that would do it.
18:15:56 <lmacken> as for projects disappearing, I have no idea how that happened to
begin with, but it has yet to happen again
18:16:19 <nirik> #info bodhi2 is in jenkins now and running tests on commits
18:16:21 <oddshocks> jenkins is just trying to get out of work. slacker.
18:16:22 <pingou> yeah the plugins need to be installed via the ansible playbook
18:16:52 <pingou> not via jenkins' UI (or done in both place, the playbook in
addition to the UI)
18:17:30 <nirik> ok. I think relrod setup the fedmsg plugin there? relrod and
pingou: can you see if you can get that in ansible (if it's not)
18:17:56 <relrod> yeah, we need to ansibleize it
18:18:03 <nirik> cool
18:18:09 <pingou> relrod: there is also a problem with the yum repo that was added
18:18:11 <pingou> (the copr one)
18:18:33 <relrod> pingou: yeah, I know. Jenkins can't talk to copr for some
reason. Short term solution is put the RPM from that copr in the infra repo
18:18:36 <nirik> yeah, we talked about that...
18:18:37 <relrod> I just need to actually do it
18:18:44 * nirik nods.
18:18:49 <pingou> relrod: ok cool
18:19:16 <nirik> ok, anything else on the applications side?
18:19:43 <nirik> pingou: oh, side question: does anitya file bugzilla bugs on
things? or was that a different layer of our old cnucnu thing?
18:19:44 <relrod> one fedora mobile related thing I guess:
18:20:01 <pingou> nirik: it'll be a different layer
18:20:23 <nirik> pingou: ok, was that something tyll was running? does he know to
update? or is that something we want to run?
18:20:30 <relrod> I am working on a system for pushing out nightly updates to it via
the Google Play alpha track. This means we can lose the ugly self-updating code that is in
Mobile right now.
18:20:49 <nirik> relrod: cool.
18:20:51 <relrod> We were going to do this via Jenkins
) but ran into some issues
18:21:02 <pingou> nirik: we'll probably want a fedmsg-cnucnu as we have
18:21:09 <relrod> so it's probably going to be a separate process that just
pulls the latest APK from Jenkins and sends it to Google
18:21:12 <relrod> which is fine
18:21:20 <pingou> nirik: I'll coordinate w/ tyll
18:21:49 <nirik> relrod: ok
18:21:53 <nirik> pingou: sounds good
18:22:02 <relrod> (the issue is basically "the APK has to be signed, and we
don't really want the signing key for it on Jenkins")
18:22:52 <nirik> relrod: hum, ok... where would that live then/
18:22:53 <nirik> ?
18:23:36 <nirik> or TBD?
18:23:40 <relrod> nirik: Probably ansible private repo, and the nightly pusher can
be a cloud node or something? Doesn't need to be very powerful
18:23:47 <nirik> ok
18:24:23 <relrod> If we don't want it in the ansible private repo, open to
18:24:40 <nirik> #action pingou to coordinate with tyll on cnucnu bugzilla filing
under the new anitya setup.
18:24:47 <nirik> I think that would be ok
18:25:05 <nirik> #info fedora mobile getting setup to be updated from the google
play alpha track.
18:25:11 <relrod> I'm not sure if the alpha signing key can be different than
the production track signing key (if anyone has Android experience and knows, poke me?
18:25:27 <nirik> no idea off hand.
18:25:39 * nirik waits for the ffos html 5 version. ;)
18:25:48 <relrod> :P
18:25:56 <relrod> anyway, that's all I have.
18:26:00 <nirik> cool. thanks.
18:26:12 <nirik> anything else application wise? or shall I move on?
18:26:27 <nirik> #topic Sysadmin status / discussion
18:26:35 <nirik> so, we did a mass reboot earlier this week.
18:26:47 <nirik> there's a few stragglers we still need to do, but mostly
everything is done.
18:26:59 <nirik> #info mass reboot earlier this week. Most things are
18:27:18 <nirik> I migrated db04 (rhel6) to db-koji01 (rhel7)
18:27:47 <nirik> I killed db04 and keys01. We now have 0 guests at telia01
18:28:04 <smooge> <<plays taps>>
18:28:10 <pingou> I need to check what specific settings we are putting on the
postgresql.conf file we install with the postgresql_server role, as this seems to be
closely related to the memory of the host
18:28:20 <nirik> I'm also going to retire mirrorlist-serverbeach, because it
keeps having problems keeping up.
18:28:26 <pingou> by telia01
18:28:31 <pingou> bye*
18:28:45 <nirik> pingou: yeah, I think we should be more dynamic on it.
18:29:03 <nirik> pingou: also, I think we just copied our old postgresql.conf in...
we should get the rhel7 one and adjust it.
18:29:09 <pingou> make it a template and use host_vars or so
18:29:12 <nirik> because we might be missing other things the 7 one can do better
18:29:18 <pingou> also true
18:29:27 <nirik> if you want to work on that that would be great. ;)
18:29:33 <pingou> anitya-backend should have the default el7 one
18:29:41 <pingou> I can at least provide a diff :)
18:29:59 <nirik> ok
18:30:15 <nirik> #info kernel01/02 now have a bunch more memory. Thanks smooge for
getting that installed.
18:30:33 <nirik> #info db04 (rhel6) migrated to db-koji01 (rhel7)
18:30:39 <smooge> well thanks jesse. all I did was watch from afar :)
18:30:43 <nirik> #info 0 guests left at telia01
18:30:58 <pingou> nirik: how many guests left in puppet?
18:31:35 <nirik> 78
18:31:45 <pingou> k
18:31:48 <nirik> I'm going to try and move some more next week...
18:32:06 <nirik> a number are virthosts, so we will have to move all the guests,
update, move back
18:32:23 <pingou> that'll be fun
18:32:34 <nirik> and of course we still have proxies to convert. thats 8 there
18:33:03 <nirik> but I am going to try and get there before the end of the year if
18:33:28 <nirik> #info 78 hosts left in puppet
18:33:44 <nirik> #topic nagios/alerts recap
18:33:56 * nirik digs up url. where's puiterwijk with it handy when you need him. ;)
18:35:21 <nirik> .tiny
18:35:23 <zodbot> nirik: http://tinyurl.com/pcpaya7
18:35:39 <nirik> download01.mgmt was fixed, was a IMM going wonky
18:36:08 <nirik> the datagrepper I think was the datagrepper db migration that
18:36:27 <nirik> the collab03 mail queue is due to the way that check is written. We
should fix it.
18:36:47 <nirik> it looks for anything mroe than 3 or 4 emails in queue, but thats
our list server, so sometimes it has a bunch in there it's sending out.
18:36:57 <smooge> I rebuilt bvirthost10 on the UCS cisco. much swearing and pain
18:37:02 <smooge> oops sorry
18:37:05 <nirik> thanks for that smooge
18:37:26 <nirik> mirrorlist-serverbeach is going away, so that should disappear.
18:38:01 <smooge> <<taps on lone bugle>>
18:38:15 <nirik> not sure about the bodhi02 ones... lmacken: any errors you have
seen from it lately?
18:38:26 <nirik> might have been the koji outage for the database move?
18:39:29 <nirik> anything else sysadminy?
18:40:02 <nirik> #topic Upcoming Tasks/Items
18:40:03 <nirik> https://apps.fedoraproject.org/calendar/list/infrastructure/
18:40:11 <nirik> anything upcoming anyone would like to note or schedule?
18:40:33 <nirik> I'll note we go into Beta freeze 2014-10-14
18:40:45 <nirik> so do try and get anything done before then thats big/disruptive.
18:41:05 <pingou> 2 weeks from now
18:41:08 <nirik> yep.
18:41:09 <pingou> that'll be there soon
18:41:23 <nirik> I'm going to try and get some more rhel7/ansible migrations
done, but we will see
18:41:42 <nirik> might do an outage and move db01 over.
18:41:47 <nirik> will have to look.
18:41:58 <nirik> #topic Open Floor
18:42:15 <nirik> anyone have items for open floor? ideas, comments, cookie recipes?
18:42:17 <smooge> I have two cloud servers to rebuild today.
18:42:31 <smooge> and I will be out tomorrow after the EPEL meeting
18:42:55 <nirik> smooge: cool. Let me know when 09 is done and I can play with
running the setup on it.
18:43:31 <smooge> will do so
18:43:47 <smooge> next week I will be focusing on I2 and budgets
18:43:48 <nirik> If we can get that all working soon we can look at cloud migrating.
18:43:56 <smooge> or cloud
18:44:09 <lmacken> nirik: no I haven't seen any errors from bodhi02 aside from
the koji outage.
18:44:31 <nirik> I think thats mostly just a matter of telling everyone: you have
until xxxx-xx-xx to terminate your instance(s) and bring them up in the new cloud.
18:44:43 <nirik> and us doing that on the persistent ones we care about.
18:44:54 <nirik> oh, we still need to get storage working tho.
18:44:58 <nirik> lmacken: ok
18:45:09 <nirik> lmacken: just odd that bodhi02 was in the nagios alerts, but not
18:45:32 <lmacken> nirik: weird, I'll take a look
18:46:11 <nirik> thanks.
18:46:18 <nirik> ok, if nothing else, will close out in a minute or so...
18:47:41 <nirik> Thanks for coming everyone! Lets all continue in #fedora-admin,
#fedora-apps and #fedora-noc.
18:47:43 <nirik> #endmeeting