=== mrevell is now known as mrevell-lunch === mrevell-lunch is now known as mrevell [17:14] premeeting tailgate [17:17] a pre-gate? [17:17] head gate? [17:17] ah, no, a jump start! [17:18] seems we've lost mootbot for today. I've pinged the scribes team but I doubt it will be fixed in time [17:19] what happened? [17:37] yea! Welcome back MootBot [17:41] Rinchen: bet you're happy about that [17:42] indeed I am! [17:46] woohoo bot is back [17:54] welcome welcome. There are plenty of chairs. [17:55] There are a few down here in here front [17:55] now, let's see if this project works [17:55] #startmeeting [17:55] Meeting started at 17:55. The chair is Rinchen. [17:55] Commands Available: [TOPIC], [IDEA], [ACTION], [AGREED], [LINK], [VOTE] [17:56] Welcome to the Launchpad Developers Meeting! [17:56] The purpose of the meetings is to coordinate Launchpad development, and particularly to ensure that none of the developers is blocked from doing what they need to do. [17:56] [TOPIC] Agenda [17:56] New Topic: Agenda [17:56] * [17:56] Roll call [17:56] * [17:56] Agenda [17:56] * [17:56] Next meeting [17:56] * [17:56] Actions from last meeting [17:56] * [17:56] Oops report (Matsubara) [17:56] * [17:56] Critical Bugs (Rinchen) [17:56] * [17:56] Bug tags [17:57] * [17:57] Operations report (mthaddon) [17:57] * [17:57] DBA report (stub) [17:57] * [17:57] Sysadmin requests (Rinchen) [17:57] * [17:57] New packages required (salgado) [17:57] * [17:57] A top user-affecting issue (mrevell) [17:57] ug [17:57] Blockers [17:57] Thumper just pointed out that my clock is apparently off [17:57] Only four minutes or so... [17:57] I was just sobering up [17:57] heh :) [17:58] lol [17:58] Rinchen: dress rehearsal? [17:58] apparently [17:58] * Rinchen kicks ntp [17:58] since when are we having launchpad meetings dressed?... [17:58] heh [17:58] * sinzu1 covers up === sinzu1 is now known as sinzui [17:59] well, let's do an early [17:59] [TOPIC] Roll Call [17:59] New Topic: Roll Call [17:59] me [17:59] me [17:59] me [17:59] me [17:59] me [17:59] me === EdwinGru is now known as EdwinGrubbs [17:59] me [17:59] me [17:59] me [17:59] me [17:59] me [17:59] me [17:59] me [17:59] me [17:59] me [17:59] me [17:59] me [17:59] [ACTION] Rinchen to fix intro script on the meeting page [17:59] ACTION received: Rinchen to fix intro script on the meeting page [18:00] me [18:00] am i late? [18:00] mr [18:00] me [18:00] me [18:00] flacoste, early [18:00] [ACTION] Rinchen to fix agenda so it's cut-n-paste-able [18:00] ACTION received: Rinchen to fix agenda so it's cut-n-paste-able [18:00] me [18:00] ACTION Rincen to fix his ntp setup [18:00] stub: indeed! [18:00] yeah, my ntp is hosed. [18:01] me? [18:01] stub: you missed the [] [18:01] :) [18:01] Rinchen: the agenda is actually quite cut-n-paste friendly if you use raw mode :-) [18:01] Rinchen jumped the gun [18:01] I'm doing my normal...let's start early routine apparently [18:01] Anyone who hasn't said ME, please do so now [18:01] said ME! [18:01] ME [18:02] thanks [18:02] [TOPIC] Next meeting [18:02] New Topic: Next meeting [18:02] The next meeting is on the 14th. [18:02] kiko, same time, 18:00? [18:02] ME [18:02] I like this time more [18:02] Rinchen: maybe 17:58? [18:02] yes! [18:03] [AGREED] Next meeting 14 Feb @ 18:00 UTC [18:03] AGREED received: Next meeting 14 Feb @ 18:00 UTC [18:03] [TOPIC] Actions from last meeting [18:03] well, i /know/ one of these days i'll take a late lunch and forget :) [18:03] New Topic: Actions from last meeting [18:03] me [18:03] barry: get google calendar to remind you :) [18:03] there were no actions recorded [18:03] thumper: doesn't help when you're not in front of your lappie [18:04] barry: it can SMS you [18:04] [TOPIC] Oops report (Matsubara) [18:04] New Topic: Oops report (Matsubara) [18:04] * sinzui 's phone has the launchpad cal on it [18:04] * barry has cheap-ass cell services :) [18:04] jtv: Can you take a look at OOPS-764F1198? I think it's related to bug 186548, but didn't happen during an import [18:04] Launchpad bug 186548 in rosetta "Still an "iterable argument required" on Slovenian translation import" [Medium,New] https://launchpad.net/bugs/186548 [18:04] https://devpad.canonical.com/~jamesh/oops.cgi/764F1198 [18:05] flacoste, the meeting is [18:05] hoser! [18:05] intellectronica: the +roadmap timeout bug is targeted to this cycle, but has no assignee. It's currently the second top timeout of the week [18:05] intellectronica: bug 185135 [18:05] Launchpad bug 185135 in blueprint "+roadmap page still times out" [Undecided,Confirmed] https://launchpad.net/bugs/185135 [18:05] * Aloha likes the 1800 time [18:05] matsubara: I was supposed to look at 186548, didn't have the time this past week [18:05] matsubara, I saw a couple of odd oopses in the latest reports -- have you gone through them? [18:05] matsubara: i'll take a look [18:05] ah! also [18:05] danilos: right. is the OOPS I pasted above related to it? [18:06] I have found a number of 404s related to merged accounts [18:06] salgado, have some time to look over them with me later today? [18:06] matsubara: very much looks like it; how many of those do we get? [18:06] [ACTION] intellectronica to investigate Bug 185135 [18:06] ACTION received: intellectronica to investigate Bug 185135 [18:06] matsubara: certainly looks related [18:06] matsubara: we used to have a single occurrence before, which is why I didn't bother [18:06] kiko: I changed my username and now I get 404s. [18:06] kiko: I think we linkify a $name-merged in some portlet still [18:06] kiko, could it be tomorrow? I am on call today and have a call with flacoste after the meeting [18:06] there's a bug for open for that [18:06] It would be nice to have redirects. [18:06] salgado, okay, tomorrow then, please remind me [18:07] abentley: the point of merging is to also free up the name space [18:07] matsubara, OOPS-763C1027, OOPS-763D1843 OOPS-767B2435 OOPS-767E2288 [18:07] https://devpad.canonical.com/~jamesh/oops.cgi/763C1027 [18:07] https://devpad.canonical.com/~jamesh/oops.cgi/763D1843 [18:07] https://devpad.canonical.com/~jamesh/oops.cgi/767B2435 [18:07] https://devpad.canonical.com/~jamesh/oops.cgi/767E2288 [18:07] [ACTION] Salgado and Kiko to discuss recent 404s [18:07] ACTION received: Salgado and Kiko to discuss recent 404s [18:07] there's another oops in the DataTimeWidget which I haven't reported yet. will do after the meeting [18:07] did someone from Translations take the early oops request? [18:07] jtv? [18:08] flacoste: sure, but until the freed name is taken, you can redirect to the appropriate place. A grace period. [18:08] Rinchen: jtv did [18:08] [ACTION] jtv to investigate OOPS-764F1198 [18:08] https://devpad.canonical.com/~jamesh/oops.cgi/764F1198 [18:08] ACTION received: jtv to investigate OOPS-764F1198 [18:08] https://devpad.canonical.com/~jamesh/oops.cgi/764F1198 [18:08] me [18:08] * Rinchen waves to adeuring [18:08] Rinchen: I already looked at it, it seems to be instance of the bug matsubara mentioned [18:08] Rinchen: that oops is related to the bug danilo will check out [18:09] me! [18:09] [ACTION] matsubara to file a bug about the DateTimeWidget oops (https://devpad.canonical.com/~jamesh/oops.cgi/OOPS-765EB80) [18:09] ACTION] matsubara to file a bug about the DateTimeWidget oops (https://devpad.canonical.com/~jamesh/oops.cgi/OOPS-765EB80) [18:09] [ACTION] matsubara to file a bug about the DateTimeWidget oops (https://devpad.canonical.com/~jamesh/oops.cgi/OOPS-765EB80) [18:09] ACTION received: matsubara to file a bug about the DateTimeWidget oops (https://devpad.canonical.com/~jamesh/oops.cgi/OOPS-765EB80) [18:09] thanks Rinchen [18:09] meeting chair permissions, sorry [18:09] anything else on oopses? [18:10] I think I'm done for now. thanks [18:10] hey! [18:10] great thanks matsubara [18:10] what about the oopses I posted matsubara? [18:10] kiko: isn't that the 404 you and salgado are going to investigate? [18:10] kiko: The last one was a known problem with tar file uploads (from what I see in the source, malformed ones) [18:11] matsubara, no, none of them are 404s. [18:12] [ACTION] matsubara to investigate OOPS-763C1027, OOPS-763D1843 OOPS-767B2435 OOPS-767E2288 [18:12] https://devpad.canonical.com/~jamesh/oops.cgi/763C1027 [18:12] ACTION received: matsubara to investigate OOPS-763C1027, OOPS-763D1843 OOPS-767B2435 OOPS-767E2288 [18:12] https://devpad.canonical.com/~jamesh/oops.cgi/763D1843 [18:12] https://devpad.canonical.com/~jamesh/oops.cgi/767B2435 [18:12] https://devpad.canonical.com/~jamesh/oops.cgi/767E2288 [18:12] https://devpad.canonical.com/~jamesh/oops.cgi/763C1027 [18:12] https://devpad.canonical.com/~jamesh/oops.cgi/763D1843 [18:12] https://devpad.canonical.com/~jamesh/oops.cgi/767B2435 [18:12] https://devpad.canonical.com/~jamesh/oops.cgi/767E2288 [18:12] moving on... [18:12] [TOPIC] Critical Bugs (Rinchen) [18:12] New Topic: Critical Bugs (Rinchen) [18:12] I want to discuss https://devpad.canonical.com/~matsubara/oops.cgi/2008-01-30/B1498 with someone [18:12] * kiko kicks Rinchen [18:12] later [18:12] Hi, I have no critical bugs to talk about today but I would like to talk about bug which causes the "choose" widget to fail for beta team members. Thanks to sinzui for doing the initial problem determination. I wanted a consensus that this is really high and not critical. [18:12] [LINK] https://bugs.edge.launchpad.net/launchpad/+bug/189742 [18:12] LINK received: https://bugs.edge.launchpad.net/launchpad/+bug/189742 [18:13] I have set that as high and not critical. Does anyone have any issues with that? [18:13] * thumper looks [18:13] does it actually cause something to not work? [18:13] I'm kind of surprised at that [18:14] Yes. Clicking on it does nothing. [18:14] I'd expect a JS error to be logged [18:14] ok [18:14] SteveA: I remember jtv looking into it and suspecting some sort of cache issue [18:14] no issues, but it's weird [18:14] SteveA: The popup widget will fail [18:14] cache issue?! [18:14] nah [18:14] it's just going to the wrong host [18:14] SteveA: (the OOPS for 'TranslationMessage deleted') [18:14] no problem -- we can fix it in apache easily enough [18:14] danilos: the cache issue wasn't me [18:15] We're agreed to leaving it as high then. Objections? [18:15] Rinchen: I'd propose doing a workaround in apache then doing a proper fix as a medium priority task [18:15] agreed [18:15] jtv: didn't you mention something about looking into https://devpad.canonical.com/~matsubara/oops.cgi/2008-01-30/B1498 already? [18:15] danilos: yes, but the cache issue wasn't me [18:15] [AGREED] work around should be done in apache and the proper fix can be a medium task [18:15] AGREED received: work around should be done in apache and the proper fix can be a medium task [18:16] [ACTION] Rinchen to update the bug report [18:16] ACTION received: Rinchen to update the bug report [18:16] jtv: you are totally confusing me [18:16] any further discussion on critical bugs before we move on? [18:16] danilos: I did not suspect a cache issue. [18:16] jtv: ok [18:16] [TOPIC] Bug tags [18:16] New Topic: Bug tags [18:17] We have no proposed bug tags [18:17] just a sec [18:17] to clarify [18:17] the cache issue discussion was about OOPS B1498 [18:17] not about the +dims on edge issue [18:17] clarification over [18:17] [TOPIC] Operations report (mthaddon) [18:17] New Topic: Operations report (mthaddon) [18:17] Nothing major to report, rather disappointingly [18:18] unless there are any questions, back to you, Rinchen [18:18] [TOPIC] DBA report (stub) [18:18] New Topic: DBA report (stub) [18:18] DB review review call with Mark scheduled for Monday. I don't forsee any issues with this months patches so hopefully formal reviews will all be done Monday or Tuesday at the latest. [18:18] Purchasing new kick arse db hardware to replace the existing kick arse db hardware is in the final stages. We can expect a little downtime as things are rsynced across. Thanks to elmo for the benchmarking and driving this through over the xmas break. [18:18] https://launchpad.canonical.com/AuthPersonSplit isn't finished or polished but is ready enough for people to have a look and approve or scream. [18:18] Me done. [18:19] question [18:19] stub: does this mean that the db won't be open until Tuesday? [18:19] thumper: Given it is Friday, it can open as soon as someone lands a branch that opens it. [18:20] thumper: If there is something approved from a previous cycle go for it. [18:20] stub, its friday for you? [18:20] Aloha: and me [18:20] 1:20am [18:20] and me [18:20] woah its thursday morning here [18:20] stub: I'll talk to you later about my patch [18:20] I propose a call sometime next week with me, stub and some people from the Foundations team to talk over AuthPersonSplit [18:20] good idea [18:21] It will need discussion. [18:21] Aloha: 7:21am Friday here [18:21] [ACTION] Steve to arrange a call sometime next week with stub and some people from the Foundations team to talk over AuthPersonSplit [18:21] thumper, almost 24 hours difference [18:21] ACTION received: Steve to arrange a call sometime next week with stub and some people from the Foundations team to talk over AuthPersonSplit [18:21] Anything else on DB? [18:22] hmm my ntp client is fixed [18:22] [TOPIC] Sysadmin requests (Rinchen) [18:22] New Topic: Sysadmin requests (Rinchen) [18:22] I have nothing to report on this, this week. Is anyone blocked on an RT or have any that are becoming urgent? [18:22] not me [18:22] I will add that since lamont has joined IS, we have seen in an increase in ticket closures [18:22] I posted an RT but forgot about it [18:23] flacoste, barry - are the mailman RTs still ok as is? [18:23] Rinchen: i think so [18:23] Rinchen: I wouldn't say it was urgent but it'd be nice to get the Launchpad News blog some spam filtering. I'm hunting for the RT number. [18:24] I saw those 258 awaiting moderation comments [18:24] Rinchen: Yeah, and because one of the spams is huge, WP freaks out and prevents moderation [18:24] [ACTION] mrevell to ping Rinchen about new blog spam [18:24] ACTION received: mrevell to ping Rinchen about new blog spam [18:24] thanks [18:25] [TOPIC] New packages required (salgado) [18:25] New Topic: New packages required (salgado) [18:25] if any of the features you have targeted to this cycle depend on any library which is not in launchpad-dependencies or inside our sourcecode/ directory, come talk to me [18:25] packages for lp? [18:25] that's not really a new topic [18:25] salgado, I have one [18:25] lp dependencies [18:25] salgado: turbogears for codebrowse [18:25] salgado, cjwatson pointed out germinate is missing [18:26] any others? [18:26] * thumper wonders what germinate is [18:26] are we getting launchpad-dependencies inside our sourcecode tree? [18:26] I'd like that [18:26] thumper: it's a tool to work out the packages that appear in a distro [18:26] [ACTION] salgado to investigate adding turbogears for codebrowse and germinate [18:26] ACTION received: salgado to investigate adding turbogears for codebrowse and germinate [18:26] thumper: from a set of initial seeds, then looking at their dependencies [18:26] * thumper nods [18:27] Rinchen, I'm done, btw [18:27] waiting on your reply salgado to Steve's question [18:27] Rinchen: i need to discuss this with salgado [18:27] are we getting launchpad-dependencies inside our sourcecode tree? [18:27] k [18:28] Rinchen: it was on my todo list, but forget about it during the sprint [18:28] [ACTION] flacoste to discussing adding launchpad-dependencies inside the sourcecode tree with salgado [18:28] ACTION received: flacoste to discussing adding launchpad-dependencies inside the sourcecode tree with salgado [18:28] no problem, thanks [18:28] [TOPIC] A top user-affecting issue (mrevell) [18:28] New Topic: A top user-affecting issue (mrevell) [18:28] Hello. I don't have any user-affecting issues this week as all the usual sources have either dealt with issues I've raised previously or have been quiet. [18:28] I'm concerned with rosetta timeouts again [18:29] how about we resurrect the ole ajax suggestions patch? [18:29] it's a simpler change and would work in the shorter term [18:29] than a big refactoring [18:29] opinions? [18:29] kiko: as I said last time you said that (I think it was last week) I prefer that we profile the code before going for ajax... [18:29] kiko: I'd still prefer going for the fixing only those messages which cause timeouts [18:30] after having some profile information, we will know whether ajax is needed or trivial changes would improve the performance [18:30] aren't we sure we're timing out because of the multiple requests for suggestions? [18:30] kiko: we aren't [18:30] okay. where's the profile, then? [18:30] whats the address for LP news blog? [18:30] I've been pushing strongly for suggestions to be done async [18:30] Aloha: news.launchpad.net [18:30] there is no profile in the usual sense: we have oops reports. [18:31] that isn't a very straight answer [18:31] kiko: I am pretty sure we are timing out because of the few very common messages and their external suggestions [18:31] mrevell, thnx [18:31] danilos: do you have documented your last profile work? [18:31] can someone do the investigation and conclude and make a plan to fix this by next week? [18:31] I just want a plan, not a fix [18:31] carlos: that was before the refactoring [18:31] kiko: sure, I'll do that [18:31] danilos: I'm not talking about the data you got but the procedure you followed to do the profiling [18:31] * kiko nudges Rinchen [18:32] was that action for danilos and the profile or the whole time out issue? [18:32] carlos: I don't have it documented, it's not really a big deal [18:32] List of possible ideas here: https://launchpad.canonical.com/Translations/Timeouts [18:32] Rinchen, yes [18:32] Rinchen: danilo does the profile, and the whole team prepares a proposal for next meeting [18:33] based on the data we got [18:33] [ACTION] danilos to investigate profiling and Rosetta team to prepare proposal for next week to fix remaining timeouts [18:33] ACTION received: danilos to investigate profiling and Rosetta team to prepare proposal for next week to fix remaining timeouts [18:33] matsubara: can I get a list of all the timeouts for +translate pages (I need exact message IDs where fetching external suggestion fails)? [18:33] danilos: it varies. [18:33] matsubara: I mean, is it possible to get that? (for the last week or so) [18:33] I see no special request topics.... does anyone have any last minute items.. steve kiko? [18:33] jtv: I know it varies [18:34] I wonder if there's a way to get the data by analyzing our oops reports specially... [18:34] danilos: I mean, once it's loaded, it stops timing out. [18:34] danilos: yes, I can get them for you. [18:34] pas mois [18:34] jtv: not necessarily, I see many times same page timing out for 37 times or so [18:34] matsubara: cool, thanks, should I ping you about it tomorrow? [18:34] danilos: until it's fully loaded, yes. [18:34] danilos: yes, please do. [18:34] matsubara: cool, thanks [18:35] [TOPIC] Blockers [18:35] New Topic: Blockers [18:35] [AGREED] Releases Team: Rinchen blocked on Moin OpenID nickname decision. Rinchen blocked on team support in Moin. [18:35] AGREED received: Releases Team: Rinchen blocked on Moin OpenID nickname decision. Rinchen blocked on team support in Moin. [18:35] Foundations: not blocked [18:35] Translations team: Not blocked [18:35] Code: not blocked [18:35] Bugs: not blocked [18:35] Rinchen, what's this openid nickname stuff? [18:35] I have read about it in emails but forgot all about it [18:35] Rinchen: agreed? [18:35] carlos, yeah so it shows up in the summary [18:35] Rinchen: [ACTION] matsubara to get last week +translate timeouts OOPS ids for danilo [18:35] kiko: OpenID returns the launchpad nickname instead of the wiki name [18:36] [ACTION] matsubara to get last week +translate timeouts OOPS ids for danilo [18:36] ACTION received: matsubara to get last week +translate timeouts OOPS ids for danilo [18:36] Rinchen: so all blocked things should be set as agreed? [18:36] Rinchen: gracias [18:36] kiko: so on launchpad.canonical.com Rinchen is known as rinchen instead of JoeyStanford [18:36] flacoste, can OpenID not pass back more than one identifier? [18:37] I think spiv was right all along and the wikis should be using emailaddress anyway [18:37] kiko: we can pass back the appropriate wiki name based on the connecting site and the information we have in the DB [18:37] stub, that's still different from launchpad name [18:37] kiko: or we can pass extra information, but that requires the wiki to handle that extra information [18:37] (or I have to move Person.name to Account.name too) [18:37] so either way a change needs to be made [18:37] I think we should continue the wiki discussion after the meeting [18:37] flacoste, how overbooked is vednis yet? [18:37] kiko: Yes. I'm just trying to push the direction it gets changed :-) [18:37] kiko: is not [18:38] kiko, soyuz team blocked? [18:38] flacoste, we could get him to look into this since it does block a big migration [18:38] not really [18:38] statik's team blocked? [18:38] SteveA, special circumstances blocked? [18:38] hwdb: not blocked [18:38] jamesh, stub: blocked? [18:38] I'm not blocked [18:38] Nup [18:38] SC: not blocked [18:39] bigjools, cprov - you guys blocked? [18:40] I'll take that as a no. [18:40] kiko: yes, if that's what we want to do, it could [18:40] danilos, matsubara: I'd like to help you plan out how to analyze oops reports for maybe cacheing certain suggestions [18:40] Rinchen: no [18:40] kiko: but i think stub has objections to syncing wikiname data across databases [18:41] stub, only issue with using emailaddress is legacy links all over our wikis [18:41] We can sync stuff. But this is why we need the auth/person split discussion. [18:41] right [18:41] lpcomm: not blocked [18:42] thanks EdwinGrubbs [18:42] SteveA: we're just going to look into them and see if my suspicion is correct (if it is, then we're going to play with caching them) [18:42] The wikiname table needs to be refactored anyway. Great big hairy WHUI IMO [18:42] SteveA: you are welcome to help anytime, though :) [18:42] what is WHUI? [18:42] Thanks everyone for attending this week's mayhem...er Development Meeting. [18:42] flacoste: We Haven't Used It [18:42] thanks for a on time meeting [18:42] flacoste: what happens when a YAGNI isn't considered early enough [18:43] right [18:43] :) [18:43] #endmeeting [18:43] Meeting finished at 18:43. [18:43] stub, well.. we use it in all our person pages actually [18:43] what's the other one? DON? [18:43] danilos: I'm hoping we can get more specific than "we'll just look at them" :) [18:43] there are many people that have multiple wikinames [18:43] yes [18:43] thanks everyone [18:43] sub, and Rinchen is probably a big user of the feature [18:43] that's why he's so vocal about it [18:43] it was good to be here for a change [18:43] kiko: how many? [18:44] hey, i've got staging access... [18:44] i can check that [18:44] GoOD [18:44] kiko: We use the information, but the design is my issue. Maybe WHUI isn't the term, but the table design is a pain to use the information the way we want to. [18:44] SteveA: well, I can easily be more specific... I am going to count on which msgid's has the suggestion fetching been timing out, and the number of all suggestions in DB for those msgid's [18:44] stub: do a view on it, then mirror the view, perhaps? [18:44] stub, how could it be different, though? sorry, I am unimaginative today [18:45] danilos: can we talk this over on a quieter irc channel in say 15 mins? [18:45] SteveA: in my previous tests, these cases have caused the most problems (i.e. strings like "Name" which have thousands of translations per language in different PO files, thus DB needs to go through all of them) [18:45] SteveA: I am mostly planning to be out in 15 minutes === mrevell is now known as mrevell-afk [18:46] danilos: so... grab hard and soft timeout OOPSes for +translate pages... [18:47] danilos: for a particular set of days [18:47] SteveA: yeah [18:47] thumper, kiko, bug 189994, fyi [18:47] Launchpad bug 189994 in launchpad "Add python-turbogears and germinate to launchpad-dependencies" [Medium,Confirmed] https://launchpad.net/bugs/189994 [18:47] danilos: then, collate them by msgid [18:48] turbogears is fun, because of a crack headed conflicts: line in a dependency [18:48] danilos: then, work out how much time is spent in suggestion-getting queries? [18:48] SteveA: right, that's the plan (matsubara will provide me with an initial set of oopses for the last 7 days), I'll look for queries taking the most time [18:48] (turbogears depends: on python-sqlobject|python-sqlalchemy and python-sqlalchemy conflicts: with psycopg1) [18:48] kiko: We will be logging into the wikis via openid using an emailaddress, so wikiname is purely legacy for our wikis. The table design where 99.9% have 'launchpad.net' or whatever as their wikiname is a bit borked. [18:48] SteveA: exactly, and the prime suspect is the query from getExternalSuggestions [18:48] danilos: I'd like you to do this scientifically -- a scientist will write down the steps to produce a result -- in your case some figures time taken to get suggestions for particular msgids [18:49] danilos: then someone else can reproduce the analysis on other data, easily, if needed [18:49] SteveA: uhm, that's how I am going to do it [18:49] danilos: to do this, the way of doing the analysis needs to be written down [18:49] SteveA: that's how I've always did it, including the previous performance improvements I did [18:50] SteveA: and that's what I always do (except for the specific 'import cprofile' and similar bits) [18:50] cool. I'm very interested to see how it goes. I think you had a very insightful idea to cache just the most common suggestions [18:50] SteveA: sure, I'll keep you in the loop [18:50] thanks! [18:50] My rationale for moving the suggestions async as the preferred next step is that the pages would no longer time out - just some of them will take a long time to fully load but they are still usable. Even if we optimize and speed up the translation db queries, it still only *might* stop timeouts and then maybe only for a time until our dataset grows more. [18:56] stub: there's another thing we're considering to actually remove exponential growth of data we have at the moment [18:56] salgado, thanks [18:56] stub: basically, sharing potmsgset's (between potemplates in different distroserieses) [18:57] stub: of course, solutions we've discussed earlier about db replication will never hurt, and are truly the only real solution [18:57] danilos: Rosetta is still affected by load on other parts of Launchpad. Even if rosetta remains stable or shrinks it doesn't mean other parts of lp won't step up and chew up all the resources you need [18:57] stub: agreed [18:58] stub: especially with all the other considerations, like implementing search and getting stable results out of it [18:59] anyway, done for the day === kiko is now known as kiko-afk === mrevell-afk is now known as mrevell