infrastructure
LOGS
19:00:11 <nirik> #startmeeting Infrastructure (2011-06-16)
19:00:11 <zodbot> Meeting started Thu Jun 16 19:00:11 2011 UTC.  The chair is nirik. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:00:11 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
19:00:11 <nirik> #meetingname infrastructure
19:00:11 <nirik> #topic Robot Roll Call
19:00:11 <nirik> #chair goozbach smooge skvidal codeblock ricky nirik abadger1999
19:00:11 <zodbot> The meeting name has been set to 'infrastructure'
19:00:11 <zodbot> Current chairs: abadger1999 codeblock goozbach nirik ricky skvidal smooge
19:00:16 * skvidal is here
19:00:18 <nirik> who all is around?
19:00:20 * CodeBlock 
19:00:25 * ricky 
19:00:35 * athmane hello
19:01:24 * nirik waits another few for folks to wander in.
19:01:26 * abadger1999 here
19:01:35 * Flare183 is here
19:01:54 <ranjibd> present
19:01:57 * jayson_r here
19:01:59 * jsmith is here
19:02:01 * thomasj is lurking
19:02:23 <nirik> cool. Lets go ahead and dive into another fun filled meeting. ;)
19:02:31 <nirik> #topic New folks introductions and Apprentice tasks
19:02:38 <nirik> Any new folks want to say hello?
19:02:51 <nirik> Or apprentices have questions/comments/tickets/issues they would like to discuss?
19:03:13 <Flare183> I would. I am Jesse N. Richardson (aka Flare183).
19:03:22 <athmane> yes, what up with db replication
19:03:24 * jayson_r say's hi
19:03:25 <smooge> here
19:03:30 <athmane> *what's
19:03:41 <nirik> welcome Flare183
19:03:42 <skvidal> athmane: what would you like to be up with it
19:03:50 <jayson_r> Jayson Rowe here.
19:04:11 <athmane> skvidal: maybe I can help with mysql ?
19:04:12 <nirik> I think ranjibd was going to work on it, but we also had to clean out our public test instances...
19:04:38 <jayson_r> Kevin Fenzi suggested I attended yesterday. I'm a network/systems admin, and willing/able to volunteer some time to Fedora
19:04:44 <nirik> jayson_r: thats me. ;)
19:04:48 <nirik> welcome
19:05:00 <jayson_r> nirik: cool! hi!
19:05:04 <ranjibd> i did some testing around mysql replication.. which works.. thats all..
19:05:08 <nirik> ricky also expressed interest in the db replication.
19:05:17 <ranjibd> on the public test boxes
19:05:43 <jayson_r> Flare183: hey buddy - surprise seeing you here!
19:05:52 <Flare183> jayson_r: hehe :D
19:05:56 <nirik> so, perhaps ricky, ranjibd and athmane could gather sometime out of meeting and discuss next steps?
19:06:09 <ricky> Sure, I never did build that extra machine to test replication on
19:06:18 <athmane> nirik: ok
19:06:25 <ranjibd> ok
19:06:59 <nirik> it would sure be nice to not have to take everything down for reboots, and the db's are a cause of some of that.
19:07:07 <nirik> great!
19:07:17 <nirik> any other apprentice tasks to discuss? or shall we move on?
19:07:56 <nirik> as always, do ask questions in #fedora-admin/fedora-noc. ;)
19:07:59 <athmane> nirik: log stuff, epylog seems outdated (and it has a huge patch in srpm)
19:08:14 <nirik> athmane: yeah. Sadly, it's the best choice that we know of...
19:08:15 <athmane> nirik: and patch not yet marged
19:08:20 <athmane> **merged
19:08:27 <nirik> there was talk about it being more active upstream, but not sure the status.
19:08:31 <skvidal> athmane: these are all known things - but there is no other option
19:08:53 <skvidal> athmane: I talked with upstream a bit and the situation is kinda "meh"
19:08:56 <athmane> skvidal, nirik : ok
19:09:06 <skvidal> what's interesting to me is this
19:09:13 <skvidal> we cannot be the only environment dealing with this
19:09:14 <athmane> skvidal: yeah I saw the discussion
19:09:35 <nirik> yeah, does everyone write custom ones?
19:09:39 <nirik> or use non free ones?
19:10:13 <skvidal> nirik: I suspect splunk
19:10:17 <athmane> I'm used logrotate but it's very basic
19:10:20 <skvidal> is whar's happening more and more
19:10:22 <skvidal> athmane: umm
19:10:26 <skvidal> logrotate is not a log parser
19:10:26 <Flare183> Does anyone know the status of Ticket #2794?
19:10:28 <skvidal> or reporter
19:10:30 <jayson_r> nirik: we use splunk here at work
19:10:35 <ricky> .ticket 2794
19:10:36 <zodbot> ricky: #2794 (Setup rsync logs rotation/collection) - Fedora Infrastructure - Trac - https://fedorahosted.org/fedora-infrastructure/ticket/2794
19:10:41 <ranjibd> ni use new relic
19:10:45 <athmane> skvidal: yes sadly
19:10:53 <ranjibd> s/n//
19:11:00 * ricky doesn't see any updates on the ticket :-/
19:11:00 <skvidal> ranjibd: new relic?
19:11:04 <athmane> skvidal: and only send email, epylog has html output
19:11:18 <skvidal> athmane: and epylog can send mail, too
19:11:29 <athmane> skvidal: yes
19:11:38 <nirik> Flare183: LoKoMurdoK was going to work on it... if you catch him when he's around you could assist?
19:11:54 <ranjibd> skvidal: yeah.. initially we uses syslog-ng + splunk.
19:12:17 <Flare183> nirik: I should can.
19:12:30 <nirik> or add yourself to the ticket saying you would like to help. ;)
19:12:49 <nirik> so, I think we are back to seeing if we can get epylog working. ;)
19:13:03 <Flare183> nirik: I sure can.
19:13:07 <skvidal> nirik:  <sigh>
19:13:08 <nirik> sounds good.
19:13:09 <athmane> nirik: yes it seems the only option
19:13:21 <skvidal> I've got a pretty deep todo list but I've actually done epylog before - so I suspect I can do it again
19:13:34 <nirik> we could ask for an alternative... but I'm not really sure where we would ask...
19:13:48 <nirik> if it's not something we can find easily, we probibly don't want it. ;)
19:14:05 <nirik> ok, shall we move on...
19:14:06 <abadger1999> lmacken was playing around with something at  one time... but the dep tree on it was.... scarey.
19:14:22 <abadger1999> chtulu-esque scarey.
19:14:25 <athmane> nirik: ok
19:14:25 <skvidal> abadger1999: prelude-lmlm
19:14:27 <skvidal> err lml
19:14:53 <skvidal> https://dev.prelude-technologies.com/wiki/1/PreludeLml
19:15:09 <ricky> Yeah...  the "limited open source version" thing :-/
19:15:57 <nirik> #topic Upcoming Tasks/Items
19:16:07 <nirik> Just a quick dump of items I have on the radar:
19:16:10 <abadger1999> I thought it was something a guy at loggly was doing in his spare time... maybe logstash?
19:16:14 <nirik> (flood ahead)
19:16:20 <nirik> abadger1999: thats it... stash
19:16:23 <nirik> #info 2011-06-17 FPCA drop dead. (but see below)
19:16:23 <nirik> #info 2011-06-21 or so: post release housecleaning tasks.
19:16:24 <nirik> #info 2011-06-23 infra meeting
19:16:24 <nirik> #info 2011-06-24 rdu mirror mapping due
19:16:24 <nirik> #info 2011-06-28 rdu mirror changeover
19:16:25 <nirik> #info 2011-07-01 mail fi-apprentice folks.
19:16:26 <nirik> #info 2011-07-01 blogs end of life.
19:16:30 <nirik> #info 2011-07-09 Remove inactive fi-apprentice people.
19:16:32 <nirik> #info 2011-07-11 - 14: smooge and nirik at phoenix
19:16:34 <nirik> #info 2011-08-01 figure replacements for xen 03/05/09/15 (EOL 09-25 ish)
19:17:08 <nirik> anyone have things to add there? or amend? or ?
19:17:31 <CodeBlock> How's the FI calendar .ical project coming along? :)
19:17:50 <nirik> I keep meaning to do that, but get sidetracked. ;)
19:17:58 <nirik> I'll try and get that setup soon.
19:18:22 <nirik> I sent out a email about the work when smooge and I are out in phx2...
19:18:34 <nirik> please do add or mention things that you think would be good for us to do/check.
19:18:41 <nirik> and also note that we might be less available that week.
19:18:53 <smooge> looked good
19:18:56 <nirik> #topic Post release housecleaning tasks
19:18:56 <nirik> https://fedoraproject.org/wiki/Infrastructure_post_release_housekeeping
19:19:08 <nirik> I pushed this off to next week as this week we were busy with reboots and such.
19:19:09 <athmane> abadger1999: just looked, it seems java stuff  (logstash)
19:19:47 <CodeBlock> athmane: *shudder*
19:20:07 <athmane> abadger1999: sorry java+rubby
19:20:09 <nirik> so, thoughts of things to add or amend from that page welcome. I'm thinking we start looking at doing those things the 21st.
19:20:11 <athmane> **ruby
19:20:16 <abadger1999> athmane: yeah.  java and ruby and javascript
19:20:27 <skvidal> abadger1999: there's something else in there ,too
19:22:19 * abadger1999 waits
19:22:20 <nirik> we've not had java or ruby to this point... ;)
19:22:33 <abadger1999> well -- puppet is ruby.  No java though.
19:22:35 <ricky> Well we've had ruby, but not java
19:22:39 <nirik> oh true.
19:22:47 <nirik> #topic Upcoming PHX2 visit
19:22:53 * skvidal would be happy to not dep on both of those for something as critical as logging
19:22:59 <nirik> mentioned this above, but had a topic for it too...
19:23:04 <skvidal> not to mention istr it required it on every system
19:23:09 <nirik> please do add or mention things that you think would be good for us to do/check.
19:23:20 <nirik> #topic FPCA deadline
19:24:16 <nirik> abadger1999: whats the status? :)
19:24:37 <abadger1999> Summary of important stats: http://fpaste.org/L3va/
19:25:08 * CodeBlock clicks
19:25:13 <abadger1999> We still have 403 packages to be orphaned by this and 74 groups losing admins
19:25:19 <abadger1999> Full output here:
19:25:20 <nirik> so, the groups losing admins, how many of those groups would have 0 admins after that? ;)
19:25:31 <abadger1999> http://toshio.fedorapeople.org/fpca/
19:25:43 <abadger1999> Quite a few I think -- but
19:26:08 <abadger1999> I'm working on taking all the hosted groups out of that by setting them to not require the cla.
19:26:24 <ricky> Ah, good point
19:26:26 <smooge> how many critical path packages?
19:27:01 <abadger1999> That will get rid ofactually -- 44 groups losing admins... minus the hosted groups, 4 groups left.
19:27:15 <athmane> smooge: afaik, not much, maybe empathy
19:27:24 <nirik> yeah, bpepple needs to sign. ;)
19:27:50 <abadger1999> So the groups front looks to be in decent shape... I'll go through the four groups/admins there and see how big a deal it is.
19:28:03 <nirik> cool.
19:28:15 <abadger1999> packages seems to be the big decision we need to make.
19:28:48 <nirik> is there a 'packages that would be orphaned' list?
19:28:52 <abadger1999> yep
19:28:57 <nirik> or thats losing owner?
19:29:03 <abadger1999> http://toshio.fedorapeople.org/fpca/packages_losing_owner.txt
19:29:13 <lmacken> athmane, athmane: logstash is just java now (as they use jruby to spit out a single jar file)
19:29:18 <lmacken> abadger1999: ^
19:29:25 <abadger1999> lmacken: cool.
19:29:33 <athmane> lmacken: thanks for the info
19:29:37 <abadger1999> so deps are more sane.
19:29:55 <lmacken> yeah, the end result is "sane"... not production ready IMO, as it doesn't idle very well
19:30:00 <abadger1999> k
19:30:52 <abadger1999> My plan is to announce the packages that would be orphaned today on devel list.
19:31:06 <abadger1999> And if people step up, reassign packages.
19:31:15 <abadger1999> That should help there.
19:31:45 <abadger1999> But we also need to decide what our goal is before pulling the trigger.
19:31:47 <nirik> dejavu-fonts facter empathy gdm gnome-screensaver puppet xcowsay <- important packages
19:31:50 <abadger1999> Are there now?
19:32:01 <abadger1999> likely, ruby as well?
19:32:02 <athmane> nirik: fedora-easy-karma
19:32:19 <nirik> didn't see it, yeah, there are some others...
19:32:34 <nirik> in the end tho, I think important packages will get picked up
19:33:00 <athmane> nirik: sorry, gone, (maybe till is ok now)
19:33:18 <nirik> so, my proposal: announce these stats/packages, defer a week.
19:33:27 <nirik> then pull the trigger.
19:34:07 <ricky> Just curious, since you have these stats - what was the total effect of the two spammings that we did?
19:34:09 <athmane> nirik: xcowsay :)
19:34:10 <skvidal> abadger1999: quick question
19:34:12 <skvidal> those pkgs
19:34:18 <skvidal> are they also the epel versions?
19:34:19 <nirik> athmane: see... you're paying attention. ;)
19:34:32 <CodeBlock> hehe
19:34:46 <abadger1999> skvidal: Yes, it includes epel as well as fedora (with no info in the list about where the orphaning would occur)
19:34:52 <skvidal> nod
19:34:52 <ricky> Maybe people would pay attention if the subject was "[packagename] set to be orphaned"
19:35:08 <skvidal> hmm
19:35:13 <skvidal> maybe worth dropping a note to the rhel list
19:35:14 <skvidal> and centos list
19:35:27 <skvidal> about pkgs like puppet taking a long walk off of a short pier
19:35:34 <nirik> abadger1999: how hard would it be to generate an epel losing owners list?
19:35:37 <abadger1999> Okay.  Is someone here on those lists?
19:35:51 <nirik> skvidal: yeah, but it's misleading, because it has co-maintainers who will take it over.
19:35:58 <skvidal> nirik: who _might_ take it over
19:36:07 <abadger1999> nirik: Not too hard.  I can break up the existing list into two queries, one for epel, one for fedora.
19:36:29 <nirik> skvidal: Jun 14 11:13:44 <tmz>	No worries on puppet being orphaned.  I use it for work and plan to keep up with it in fedora/epel. :)
19:36:30 <abadger1999> tmz has volunteered to take over puppet I think.
19:36:40 <abadger1999> I don't know who would take over ruby, though.
19:37:02 <ricky> The maintainer for ruby is confirmed gone?  :-/
19:37:05 <skvidal> nirik: ah, okay then
19:37:11 <ricky> As in definitely not signing the FPCA?
19:37:16 <nirik> it's got 5 co-maintainers.
19:37:25 <abadger1999> ricky: No, but I pinged him and he talked about signing and still hasn't signed.
19:37:26 <nirik> ricky: he's traveling and hasn't had a chance to look at it.
19:37:29 <ricky> Ah.
19:37:31 <abadger1999> And is travelling so....
19:37:36 <ricky> That's less bad then :-)
19:37:39 <nirik> so, another week might help that as well.
19:39:11 <nirik> I don't want to wait forever, but I think another warning with the package list and more info might help people network and talk to those that still need to sign.
19:40:08 <nirik> thoughts?
19:40:29 <abadger1999> ricky: after the meeting, I need to find out from you if you think the best way to mark people's status in groups that require the cla is.
19:40:59 <ricky> Sure thing
19:41:39 <smooge> I say pull the trigger. The drama quotiant for Summer Fedora is not full yet
19:41:44 * skvidal will brb
19:41:59 <abadger1999> nirik: Sounds reasonably good.  There will be disruption from this as some people will lose their groups/packages and have to start over to regain those.
19:42:15 <abadger1999> But --- can't be helped.
19:42:35 <abadger1999> Unless you have ideas there?
19:42:59 <nirik> Not off hand. I think most groups will be somewhat accomodating to people who have this happen...
19:43:02 <abadger1999> backup the data and we manually add them back in if they sign the fpca?
19:43:05 <nirik> ie, come back and explain why they were missing, etc.
19:43:21 <abadger1999> <nod>
19:43:22 <smooge> I think in many ways too many things have been crisis driven and people won't worry about it until it really happens.
19:43:28 <ricky> We could do crazy stuff like add a new status
19:43:31 <nirik> smooge: yeah.
19:43:31 * skvidal returns
19:43:34 <smooge> But a week won't kill anything but longer than that is not helping
19:43:45 <ricky> What I was thinking was selecting out the memberships into a temporary table where we can restore them on CLA sign
19:43:53 <nirik> yeah, I am saying 1 more week, we pull the trigger after next meeting. ;)
19:44:02 <ricky> s/CLA/FPCA :-)
19:44:30 <nirik> if you have done work in, or are known in a group it shouldn't be too hard to ping a sponsor again...
19:44:41 <nirik> it might be good to have the old info for reference I suppose.
19:44:56 <nirik> $user sponsored you into $group last time... check with them....
19:45:27 <ricky> Out of curiosity, how will the pkgdb part work?
19:45:48 <smooge> abadger1999, you said earlier you are removing FPCA requirements for some groups.. will that affect say fedorapeople accounts in that if they were CLA+1 (and that was a group now not needing it).. they would need a new extra group?
19:45:49 * jayson_r waves hand
19:45:49 <ricky> Will the packages actually be manually orphaned?
19:45:55 <nirik> and for packages, many of these people are absent for a while... it would be good to get this done and people who care about the packages maintaining them.
19:45:56 <smooge> hello jayson_r
19:46:04 <jayson_r> sorry for butting in, but i have to run to a meeting - nice to meet everyone - i'll be around
19:46:05 <abadger1999> hmmm.
19:46:05 <smooge> nirik, I agree
19:46:17 <nirik> jayson_r: welcome again, and do visit us more. ;)
19:46:17 <ricky> smooge: It shouldn't, all the CLA+1 looks at is membership in the CLA group.
19:46:18 <jayson_r> nirik: you have my e-mail - let me know where i can help
19:46:19 <smooge> jayson_r, thanks. see you in #fedora-admin
19:46:28 <ricky> It doesn't care about group dependencies at all
19:46:33 <smooge> ricky, ah ok
19:46:34 <jayson_r> smooge: i'll add it ot my irssi config :-) thanks
19:46:42 <Flare183> I'll be around as well. Just ping me as needed.
19:46:43 * jayson_r says goodbye
19:46:47 * ricky is less sure about the current status of the cla_done group though.
19:47:10 <ricky> As in, does it contain fpca signers only, or cla signers as well?
19:47:12 <nirik> my understanding is that cla_done will still exist for now.
19:47:21 <ricky> But who will it contain?
19:47:25 <abadger1999> ricky: Yeah -- I may do it in the db but it'll be an orphaning of all those packages.
19:47:33 <abadger1999> cla_done will exist.
19:47:33 <nirik> cla_fpca after the trigger.
19:47:40 <nirik> and cla_redhat, etc.
19:47:40 <abadger1999> Right now it contains both cla and fpca signers.
19:47:50 <nirik> ie, remove cla_fedora from it.
19:47:51 <abadger1999> After the trigger it will not contain cla_fedora signers.
19:48:00 <ricky> OK, and cla_companyname is still all valid?
19:48:04 <abadger1999> correct.
19:48:11 <nirik> for now
19:48:26 <smooge> nirik, ricky did we upgrade the wiki this week? and how did that go?
19:48:27 <ricky> OK, cool, that's all the questions I have on this until after meeting, then :-)
19:48:32 <abadger1999> yeah, spot asked me to lok into getting rid of cla_redhat in the future.
19:48:38 <ricky> Yeah, there were a few kinks in the wiki upgrade
19:48:40 <smooge> oops sorry I was waiting for openfloor
19:48:49 <nirik> #topic week in review
19:48:50 <ricky> First was a mediawiki + lockdown bug, and second was that some CSS fixes were needed
19:48:53 <smooge> let me ask that then
19:49:04 * ricky still needs to file the bug with mediawiki.
19:49:31 <ricky> Apart from that, everything went pretty smoothly - thanks for getting the mediawiki116 package prepared
19:49:58 <ianweller> ricky: enlighten me post-meeting on the mediawiki + lockdown bug
19:50:04 <nirik> ricky: there was some request for the symantic plugin.
19:50:20 <nirik> how hard would that be to do?
19:50:25 <ianweller> semantic mediawiki?
19:50:55 * nirik looks for the plugin, yeah, that one I think. it's packaged.
19:51:05 * ricky isn't familiar with that - is it packaged specifically for mediawiki116?
19:51:12 <nirik> mediawiki-semantic
19:51:15 <ricky> Darn, it's not then.
19:51:17 <nirik> nope, seems base only
19:51:34 <nirik> could ask jlaska if he would be willing to make a 116 version
19:51:39 <ricky> It's a little unfortunate that the path changed and all :-/
19:51:39 <ianweller> i don't think it would be too difficult to adapt the plugin for 116, i could review it
19:52:03 <ianweller> ricky: well we did that because some extension versions aren't compatible with some versions of mediawiki
19:52:03 <ricky> Well, mind adding that to the ticket, and hopefully the submitter can submit the package? :-)
19:52:10 <ricky> Ah.
19:52:17 <ianweller> which ticket? :)
19:52:20 <nirik> also this last week, we did another mass rebooting/updates... this time splitting things into A, B, C groups.
19:52:28 * jlaska perks up
19:52:42 <jlaska> oh eew, semantic
19:52:52 <jlaska> I'm happy to bump that ... I didn't know anyone was using that
19:53:01 <smooge> what is useful for?
19:53:01 <ianweller> there was a request, apparently
19:53:31 <jlaska> smooge: it was something QA was exploring as a way to add more metadata options to mediawiki ... it's something OLPC used extensively for test management
19:53:57 <jlaska> the learning curve was too steep for the amount of time I was able to give it ... so it hasn't gone any further than packaging
19:54:22 <jlaska> is there a bz that I missed requesting an updated mediawiki-semantic package?
19:54:30 <nirik> liknus was wanting it for the events stiff they were doing.
19:54:40 <nirik> I don't know that a ticket has been filed yet.
19:54:43 <nirik> I can follow up on it.
19:54:53 <jlaska> I'll start looking on my end too
19:54:54 <ianweller> where was the request made? i wonder if something would better suit their needs
19:55:04 <smooge> jlaska, I don't know of a bz. We are more looking at how to deal with the mediawiki116 and mediawiki117 trees
19:55:08 <nirik> ianweller: #fedora-admin the other day. ;)
19:55:32 <nirik> I'll follow up and we can go from there.
19:55:38 <jlaska> okay
19:55:49 <nirik> #topic Open Floor
19:56:03 <ianweller> nirik: thx
19:56:04 <nirik> any other items/tickets/complains/kudos/etc anyone has?
19:56:33 <ianweller> kudos++ to ricky, smooge et al for the mediawiki upgrade!
19:56:44 <smooge> well ricky
19:56:47 <nirik> yeah, that was a great one to get done. ;)
19:56:48 <smooge> all I did was make packages
19:56:49 <jlaska> yes thank you for that!!
19:57:06 <smooge> this is our quarter for ordering lots of systems
19:57:21 <CodeBlock> :D
19:57:33 <smooge> I need to talk with nirik on what we want and get them ordered so they can be there in July
19:57:48 <nirik> smooge: yeah. We need about 5 at least ordered.
19:58:03 <nirik> we can talk out of band on it...
19:58:17 <smooge> okie dokie
19:58:21 <smooge> after meeting
19:58:22 * nirik will close out the meeting in a minute or two if nothing else comes up.
19:58:46 <skvidal> smooge: can you get the doesn't-fail hw?
19:59:02 <nirik> skvidal: I don't think we have enough budget for that. ;)
19:59:08 <skvidal> nirik: a shame, really.
19:59:25 <nirik> ok, thanks for coming everyone! as always, continue on #fedora-admin and #fedora-noc.
19:59:26 <CodeBlock> pssh, servers that work. skvidal what dream are you living in. :)
19:59:30 <nirik> #endmeeting