LPCIBot | Project parallel-test build #76: STILL FAILING in 1 hr 9 min: https://lpci.wedontsleep.org/job/parallel-test/76/ | 00:37 |
---|---|---|
LPCIBot | Project devel build #842: FAILURE in 5 hr 32 min: https://lpci.wedontsleep.org/job/devel/842/ | 02:19 |
LPCIBot | Project parallel-test build #77: STILL FAILING in 1 hr 9 min: https://lpci.wedontsleep.org/job/parallel-test/77/ | 05:10 |
LPCIBot | Project db-devel build #673: FAILURE in 5 hr 38 min: https://lpci.wedontsleep.org/job/db-devel/673/ | 06:36 |
LPCIBot | Project devel build #843: STILL FAILING in 5 hr 16 min: https://lpci.wedontsleep.org/job/devel/843/ | 07:36 |
huwshimi | A couple of UI changes to be reviewed if anyone wants to do them: | 09:09 |
huwshimi | https://code.launchpad.net/~huwshimi/launchpad/what-next-712259/+merge/66030 | 09:09 |
huwshimi | https://code.launchpad.net/~huwshimi/launchpad/table-headings-728187/+merge/66008 | 09:09 |
wgrant | bigjools: I ripped Storm out of lazr.amqp last night. | 09:10 |
bigjools | \o/ | 09:10 |
wgrant | And got it working with the rabbit fixture instead of system rabbit with hacked landscape setup script. | 09:10 |
wgrant | But it's slow. | 09:10 |
bigjools | \o/ and :( | 09:10 |
wgrant | So I need to port the test suite to testtools instead. | 09:11 |
wgrant | Rather than using zope.testrunner. | 09:11 |
wgrant | But lifeless and jml know about that. | 09:11 |
bigjools | jfdi | 09:11 |
wgrant | So it's doable this morning. | 09:11 |
wgrant | Rabbit changes aren't landed yet, but the suite passes locally with them. | 09:11 |
* StevenK giggles at "\o/ and :(" | 09:11 | |
wgrant | So you can happily turn off your system rabbit soon :) | 09:11 |
bigjools | WOOHOO! | 09:11 |
jml | wgrant: ok. I can work with you on that, since my talk is during this plenary session | 09:12 |
jml | matsubara: you should talk to flacoste & deryck about their acceptance test plans | 09:13 |
jml | matsubara: involves selenium & LEPs | 09:13 |
wgrant | jml: I think it should be pretty easy. | 09:14 |
jml | wgrant: yeah | 09:14 |
wgrant | jml: I started it yesterday but then abandoned it when I looked at docs. | 09:14 |
wgrant | I wonder how trivial a rabbit backport to lucid/maverick is. | 09:15 |
wgrant | Or possibly just skip the reject test on older rabbits. | 09:15 |
jelmer | wgrant: new deployment, w00t :) | 09:15 |
wgrant | jelmer: Yup, and the upgrades work fine. | 09:15 |
wgrant | jelmer: Not sure about the automatic Invalid stuff, though. | 09:16 |
wgrant | Thanks for fixing that up! | 09:16 |
matsubara | jml, will do. thanks for the heads up | 09:16 |
LPCIBot | Project parallel-test build #78: STILL FAILING in 1 hr 6 min: https://lpci.wedontsleep.org/job/parallel-test/78/ | 09:49 |
wgrant | There we go. | 09:55 |
wgrant | I tried persisting the rabbit process and just recreating the vhost between tests, but that seems to take 100-200ms... but it seems OK to just manually clear out the vhost over AMQP. | 09:56 |
wgrant | So I now have a fast test suite using the rabbit fixture. | 09:56 |
wgrant | yay. | 09:56 |
rvba | wgrant: Could you please update DF when you get a chance? | 10:03 |
StevenK | Wait until I'm done with it. | 10:05 |
StevenK | Please. :-) | 10:05 |
matsubara | Could someone review this branch for me: https://code.launchpad.net/~matsubara/launchpad/724727-single-line-inline-editor/+merge/66089? It's a small css change | 10:11 |
wgrant | StevenK: (too late) | 10:14 |
wgrant | But I didn't restart the appserver, since I know you're using the librarian. | 10:14 |
StevenK | Right | 10:14 |
StevenK | Can you prod and see if memcached is running? | 10:14 |
wgrant | There is a memcached running. | 10:14 |
wgrant | (zomg, no latency!) | 10:15 |
StevenK | 2011-06-28 09:15:04 WARNING Memcache set failed for populate-bprc | 10:15 |
StevenK | :-( | 10:15 |
wgrant | Checking conifg.s | 10:15 |
wgrant | ah. | 10:15 |
wgrant | StevenK: The system memcached is running, but the Launchpad one is not. | 10:16 |
wgrant | Should I start it? (needs appsever/librarian restart) | 10:16 |
StevenK | Wait until garbo finishes | 10:16 |
wgrant | k | 10:17 |
StevenK | 2011-06-28 09:17:56 DEBUG2 [PopulateBinaryPackageReleaseContents] Iteration 1106 (size 1.0): 3.701 seconds | 10:18 |
lifeless | morning guys | 10:22 |
wgrant | Good day sir. | 10:22 |
lifeless | I have my webcam ready to roll | 10:22 |
lifeless | if someone can skype me that would be great | 10:22 |
wgrant | Excellent. | 10:22 |
StevenK | 2011-06-28 09:23:24 DEBUG2 [PopulateBinaryPackageReleaseContents] Done. 1192 items in 1192 iterations, 3267.929582 seconds, average size 1.000000 (0.364756941707/s) | 10:23 |
StevenK | wgrant: Please to be restarting. | 10:23 |
lifeless | spm: http://onefte.com/2011/06/27/youre-such-an-enabler/ is the bomb! | 10:25 |
wgrant | StevenK: Should be running now. | 10:29 |
=== almaisan-away is now known as al-maisan | ||
=== al-maisan is now known as almaisan-away | ||
=== almaisan-away is now known as al-maisan | ||
jml | lifeless: technology ensues | 10:33 |
lifeless | \o. | 10:34 |
=== al-maisan is now known as almaisan-away | ||
jml | lifeless: inappropriate branding! | 10:34 |
lifeless | jml: save-as branding | 10:35 |
StevenK | lifeless: Can we troll you instead? | 10:42 |
StevenK | lifeless: Seriously, please talk a little slower. | 10:42 |
wgrant | It's not that bad... | 10:42 |
lifeless | call dropped | 10:43 |
lifeless | or held ? | 10:43 |
wgrant | Francis killed it. | 10:43 |
wgrant | Somehow. | 10:43 |
jml | lifeless: flacoste pressed the wrong button | 10:43 |
jml | I bet he hit <space> | 10:43 |
bigjools | PEBKAC | 10:43 |
=== mrevell_ is now known as mrevell | ||
jml | lifeless: that means "99%ile is twice as fast"? | 10:47 |
jml | https://dev.launchpad.net/ArchitectureGuide/Services | 10:51 |
StevenK | lifeless: Does it suffer from the /tmp issue? | 10:54 |
StevenK | \o/ | 10:54 |
poolie | lifeless: there's going to be more than 86 lines of work in getting it deployed, is my impression | 10:55 |
jml | poolie: hah! | 10:55 |
poolie | for example the requirement for monitoring etc | 10:55 |
poolie | lifeless: i like the way you've kept banging the drum of performance tuesday | 10:55 |
poolie | lifeless: it might be easier to put questions on irc | 10:56 |
poolie | jml: ? | 10:56 |
jml | lifeless: <wallyworld> we've chosen web.py as our web infrastructure for our microservices, we've had other suggestions here that might be better (e.g. pyramid), why have we chosen web.py? | 10:57 |
jml | poolie: the "more than 85 lines of code" comment | 10:57 |
jml | poolie: it's true. | 10:57 |
poolie | ah | 10:58 |
poolie | i hoped for a moment you were laughing because i was wrong :) | 10:58 |
wgrant | poolie: Well, I think that initial cost is high, but it will become tiny once we have a few services set up similalr.y | 10:58 |
jml | wgrant: yeah, that's what I was thinking | 10:58 |
jml | poolie: maybe one day you will be :) | 10:59 |
poolie | yeah, and as IS continue their projects for puppet, or as things become more containerized | 10:59 |
wgrant | Like haproxy is now. | 10:59 |
wgrant | Initial cost was high. | 10:59 |
wgrant | But now we have it set up, we can add new stuff to it trivially. | 10:59 |
jml | also as we figure out our monitoring needs & come up with stuff that is easy to use, for example | 10:59 |
wgrant | Right. | 10:59 |
poolie | lifeless: this seems a bit inconsistent with saying that we can't use graphite because of (what was it) mental load | 11:00 |
wgrant | It's going to be a slow start. | 11:00 |
jml | although very much agree w/ what lifeless is saying right now about not converging on tech choices too soon. | 11:00 |
poolie | me too | 11:00 |
jml | what are you banging? | 11:01 |
wgrant | I think pyramid is similarly inappropriate. | 11:01 |
wgrant | It seems to be an application framework. | 11:01 |
wgrant | Smaller, but still an application framework, sort of. | 11:01 |
jml | https://dev.launchpad.net/ArchitectureGuide/ServicesRequirements | 11:02 |
jml | for those who want to read them | 11:02 |
poolie | i do agree with experimenting with new technologies, and that new ones are needed | 11:02 |
wgrant | Thanks jml | 11:02 |
stub | I was considering hacking up a version of Robert's work to use ZeroMQ so a direct comparison can be made. | 11:03 |
wgrant | lifeless: But don't the LOSAs already manage graphite instances, so it's not quite the same thing? | 11:04 |
poolie | i guess i find it hard to believe having two graphing services will cause more mental load than having some code in node and same in python | 11:04 |
poolie | nm | 11:04 |
poolie | i don't want to side track | 11:04 |
poolie | the general issue is: what's the amount of diversity that is acceptable? | 11:05 |
poolie | any amount? | 11:05 |
poolie | i'm not trying to debate the particular case here | 11:05 |
jml | so the problem is that it would violate the need for a canonical, trusted, operational data source | 11:05 |
jml | rather than that it would be new tech? | 11:06 |
poolie | lifeless: we're hanging up on you | 11:09 |
jml | lifeless: you're still on screen, fwiw | 11:09 |
poolie | lifeless: i don't mind changing this to tuolumne (much) but i don't understand the architectural position behind it | 11:10 |
poolie | well... i can understand saying it's easier to do them in two steps | 11:10 |
lifeless | poolie: graphing is a core realtime troubleshooting service | 11:11 |
lifeless | poolie: development of a service is a nonrealtime isolated service | 11:12 |
lifeless | poolie: it isn't an architectural position to say that the troubleshooting stuff - logs, graphs, etc - need to be as consistent and reliable as we can | 11:12 |
lifeless | poolie: its a pure ops position | 11:13 |
poolie | don't frameworks have big ops consequences? | 11:13 |
poolie | well, perhaps microframeworks don't, but deploying say go or nodejs seems likely to substantially impact operational debugging | 11:14 |
poolie | well | 11:14 |
lifeless | poolie: there will be an impact yes; its not free | 11:14 |
lifeless | poolie: but by their nature they are more isolated than graphing | 11:15 |
lifeless | bigjools: so, around | 11:16 |
lifeless | ? | 11:16 |
poolie | it's a coffee break now | 11:22 |
poolie | i' guess he'll be back later | 11:22 |
lifeless | poolie: another way to talk about the tuolumne/graphite thing | 11:24 |
lifeless | poolie: for a given microservice, we'll only have *one* implementation live | 11:25 |
lifeless | poolie: (unless we're deliberately migrating [as a funded driven project, not an itch-scratch that could stop at any time] betweeen implementations) | 11:25 |
poolie | that's a reasonable point | 11:29 |
poolie | sinzui, when is that bugfix likely to go out onto qas? | 11:30 |
lifeless | poolie: I can tell you are confused; you feel its inconsistent : I don't think it is, but IRC late at night isn't a brilliant way to tease it out. if you want voice, I can happily do that. | 11:31 |
poolie | it's not a practical problem for me at the moment | 11:32 |
poolie | i'm sorry it apparently took over your session, which wasn't my intention | 11:32 |
lifeless | no worries; it was only a slight segue | 11:33 |
poolie | could someone send me the page performance report url? | 11:33 |
lifeless | poolie: https://devpad.canonical.com/~lpqateam/ppr/lpnet/latest-daily-partition.html | 11:33 |
poolie | the stats stuff i've been reading reinforced to me that its treatment of times as normally distributed is very bogus | 11:34 |
lifeless | hugely | 11:34 |
lifeless | patches solicited! | 11:34 |
poolie | i like the way it has highly optimized algorithms from Knuth implementing the wrong solution | 11:34 |
poolie | complete with citations of learned papers | 11:34 |
poolie | anyhow, go and sleep | 11:35 |
lifeless | poolie: well, for the purpose of giving us -a- baseline and running ok, its fit for purpose | 11:35 |
poolie | right | 11:35 |
poolie | as it speeds up the numbers will go down so it may be mostly academic | 11:36 |
lifeless | poolie: I would be worried if I thought the reported 99th percentile would be low. | 11:37 |
lifeless | poolie: I don't think it will be | 11:37 |
matsubara | benji, https://code.launchpad.net/~matsubara/launchpad/724727-single-line-inline-editor/+merge/66089 | 11:38 |
lifeless | bigjools: ping | 11:39 |
bigjools | lifeless: you punged? | 11:39 |
lifeless | I did | 11:39 |
lifeless | do you have a few minutes for a private skype call ? | 11:39 |
bigjools | lifeless: for you, of course, let me get set up, one min | 11:40 |
bigjools | lifeless: actually, 5 mins, I need to run to my room to get my headset | 11:40 |
lifeless | thats fine | 11:40 |
lifeless | I shall go talk to lynne for a minute ;) | 11:40 |
bigjools | lifeless: ok one minute again, allenap lent me his | 11:41 |
huwshimi | Anyone want to do some ui reviews for me? | 11:43 |
huwshimi | More specifically: https://code.launchpad.net/~huwshimi/launchpad/what-next-712259/+merge/66030 and https://code.launchpad.net/~huwshimi/launchpad/table-headings-728187/+merge/66008 | 11:43 |
bigjools | lifeless: hang on | 11:43 |
bigjools | headset fail | 11:43 |
lifeless | bigjools: hanging :> | 11:43 |
bigjools | as in skype doesn't want to use it | 11:44 |
lifeless | poolie: deployment overhead for microservices needs automation, but they should be (roughly) one init script [cookie cutter], one log feeder into tuolumne (standard apache format, so no coding), one nagios probe to hit a good url (also cookie cutter), and a stock haproxy config (also cookie cutter) | 11:46 |
lifeless | jml: I haven't written up the testing/importing aspects of services yet, today was more fragmented than I anticipated | 11:47 |
lifeless | stub: a 0mq service impl would be interesting indeed | 11:48 |
lifeless | stub: I am not convinced by my own analysis about ha needs yet ;) - having some experience and learning there would be most excellent | 11:49 |
poolie | i wonder about the 99th percentile; it seems it ought to be possible to calculate it mathematically | 11:50 |
poolie | i wonder if my R is up to it | 11:51 |
barry | gary_poster: when's a good time to get together with you and sinzui? | 11:51 |
lifeless | poolie: one thing to note is our dataset size - we're dealing with 12M rows per day, and for each row we have pageid, service time, sql time, sql query count | 11:52 |
poolie | so any patch needs to not break the scaling behaviour | 11:53 |
bac | has anyone seen abentley? | 11:56 |
huwshimi | bac: Just saw him walk out of the main LP room | 11:57 |
bac | thanks huwshimi...hopefully he's coming upstairs | 11:57 |
huwshimi | benji: Do you feel like doing a couple of reviews for me? | 11:59 |
jml | https://code.launchpad.net/~jml/lazr.amqp/license-and-copyright/+merge/66113 | 11:59 |
stub | lifeless: where does the updated batching code live? Looking at Bug #739052 with abel. | 12:00 |
_mup_ | Bug #739052: Distribution:+builds timeout <timeout> <Launchpad itself:Triaged by adeuring> < https://launchpad.net/bugs/739052 > | 12:00 |
stub | lifeless: Batching over nearly 800,000 results in a bad case | 12:01 |
stub | lifeless: nm. Found it. | 12:04 |
gary_poster | hey barry! how about right after lunch, 2-ish? | 12:04 |
barry | gary_poster: sounds good | 12:06 |
barry | gary_poster: i'll meet you in the lp room | 12:06 |
gary_poster | thanks barry! | 12:07 |
poolie | lifeless: fyi mean+3*stddev does underestimate the 99th percentile time of an exponential distribution | 12:10 |
poolie | by about 12% | 12:11 |
StevenK | lifeless: Are you around by any chance? | 12:11 |
lifeless | StevenK: yes | 12:11 |
poolie | this is based just on the generic distributions, not launchpad's specific data | 12:11 |
lifeless | stub: you're going to finish it off \o/ | 12:12 |
lifeless | ? | 12:12 |
poolie | it might actually more likely be an erlang distribution, not exponential | 12:12 |
lifeless | stub: there were only a few remaining glitches in my branch, I think. | 12:12 |
lifeless | stub: https://code.launchpad.net/~lifeless/launchpad/bug-752153/+merge/56505 | 12:13 |
StevenK | lifeless: Do you know how large the path table for the conflicts checker is? | 12:13 |
lifeless | stub: https://code.launchpad.net/~lifeless/launchpad/bug-759467/+merge/58262 is also -so-close-to-done- that will help with some timeouts | 12:13 |
lifeless | StevenK: not offhand | 12:13 |
jcsackett_ | wallyworld: some code supposedly using the App Framework i have just found. https://github.com/ericf/photosnear.me/blob/master/public/photosnearme.js | 12:14 |
poolie | so it's not collossally off, probably | 12:14 |
jcsackett_ | er, wallyworld_ ^ | 12:14 |
allenap | wgrant, gmb, jml, bigjools: The Rabbit fixture gives off working-like noises and is in lp:~allenap/+junk/lazr.fixtures | 12:16 |
allenap | rvba: Oops, you too ^ | 12:16 |
stub | lifeless: Bah. Thought that stuff all landed. | 12:17 |
jml | allenap: cool! | 12:18 |
lifeless | StevenK: the schema: | 12:19 |
lifeless | CREATE TABLE binpackagefiles (package_id INTEGER, file_id INTEGER); | 12:20 |
lifeless | CREATE TABLE filepath (id INTEGER PRIMARY KEY, path VARCHAR); | 12:20 |
lifeless | CREATE TABLE packageversion (id INTEGER PRIMARY KEY, name_id INTEGER, version VARCHAR, arch_id INTEGER, controltext VARCHAR, analyzed_level INTEGER, preinst_hash VARCHAR); | 12:20 |
lifeless | select count(*) from binpackagefiles; | 12:20 |
lifeless | 49416355 | 12:20 |
StevenK | lifeless: Right, I'm curious how large the filepath table is | 12:20 |
lifeless | its counting now | 12:20 |
lifeless | it wasn't running, so its all cold IO | 12:20 |
StevenK | ight | 12:20 |
StevenK | *Right | 12:20 |
lifeless | but clearly < 49M :P | 12:21 |
jcsackett_ | wallyworld_: https://github.com/yui/yui3/tree/master/src/app/js | 12:22 |
lifeless | StevenK: select count(*) from filepath; | 12:23 |
lifeless | 11018381 | 12:23 |
lifeless | StevenK: do you want the rows? | 12:23 |
StevenK | So 11 million or so | 12:23 |
lifeless | select * from filepath limit 1; | 12:23 |
lifeless | 1|usr/bin/2vcard | 12:23 |
StevenK | I'm trying to work out how large the BPP and BPRC tables are going to be. | 12:24 |
StevenK | BPP might only be a couple of GiB | 12:25 |
lifeless | average length 61.3 | 12:26 |
lifeless | select sum(length(path)) from filepath; | 12:26 |
lifeless | 675274107 | 12:26 |
danilos | gmb, hi, I am about to remove your JS code for "collapsibles" which I can't find being used anywhere; you've got 0.04s to complain ;) | 12:26 |
StevenK | lifeless: With 350,000 paths on DF, the table is currently 75MiB. | 12:26 |
lifeless | (e.g. < 700MB) | 12:26 |
gmb | danilos: No complaints. Kill it. | 12:26 |
danilos | gmb, cool, thanks | 12:26 |
poolie | after a branch lands to devel, how soon can i expect to see it on qas? | 12:26 |
lifeless | poolie: 7 hours | 12:27 |
benji | Riddell: http://docs.python.org/library/pdb.html#debugger-commands | 12:27 |
lifeless | poolie: up to 13 if it just missed a buildbot run | 12:27 |
LPCIBot | Project db-devel build #674: STILL FAILING in 5 hr 51 min: https://lpci.wedontsleep.org/job/db-devel/674/ | 12:27 |
lifeless | poolie: more if trunk is broken | 12:27 |
lifeless | stub: I wish they had landed :) | 12:27 |
StevenK | stub: Where are you hiding? | 12:29 |
poolie | thanks | 12:29 |
poolie | stevenk also told me to look at the estimate in builbdot and add 30m | 12:30 |
StevenK | lifeless: I question your 7. I thought qas took ~ 30 minutes to update | 12:30 |
poolie | the step up in https://lpstats.canonical.com/graphs/CodebrowseHTTPResponses/ is a bit interesting | 12:30 |
lifeless | StevenK: 6 hours to go through BB | 12:30 |
stub | StevenK: Up with Julian for rabbity stuff | 12:30 |
lifeless | StevenK: 30 minute frequency of qas updates | 12:30 |
lifeless | StevenK: 15 minutes latency for the carob copy to update after BB | 12:31 |
lifeless | StevenK: and then the build-and-deploy time on asuka | 12:31 |
stub | adeuring: I'll send lp:~lifeless/launchpad/bug-752153 off to ec2 test to see how close it is to landing. I think we need this branch to land before we can tackle the batching bug. | 12:31 |
benji | Riddell: http://docs.zope.org/zope2/zope2book/AppendixC.html | 12:32 |
adeuring | stub: ok | 12:32 |
poolie | so probably no less than 7 and possibly more? | 12:33 |
stub | lifeless: So StevenK has about 30GB of new data to store (every file path in every package). Microservice or main LP DB? | 12:34 |
lifeless | poolie: right | 12:34 |
benji | jcsackett_: did you figure anything out on bug 436247? | 12:34 |
_mup_ | Bug #436247: links in side portlets are too close <easy> <lp-bugs> <ui> <Launchpad itself:Triaged> < https://launchpad.net/bugs/436247 > | 12:34 |
jcsackett_ | benji: i wasn't able to pursue it with my computer randomly dying. :-P | 12:35 |
jcsackett_ | benji: as huwshimi pointed out, probably worth bugging sinzui | 12:36 |
=== jcsackett_ is now known as jcsackett | ||
lifeless | stub: If a microservice makes sense for it, \o/. | 12:36 |
lifeless | stub: be sure to consider the whole use case in assessing that | 12:37 |
matsubara | bigjools, is there an easy way to create a private ppa in launchpad.dev with some dummy packages in it? | 12:37 |
stub | lifeless: I was poking him to talk to you about it if you are in your teddy bear suit. | 12:37 |
lifeless | sure, I can do that | 12:37 |
lifeless | jml: can you champion my service fake tests etc today, I promise to write it up tomorrow | 12:50 |
lifeless | StevenK: so did you want to talk about doing amicroservice? skype? | 12:53 |
jml | lifeless: ok | 12:53 |
StevenK | lifeless: I haz no skype | 12:53 |
lifeless | StevenK: voip? | 12:53 |
StevenK | lifeless: If it's 30GiB, I'm comfortable doing it in the main database since it will be running as part of the publisher. | 12:54 |
lifeless | jml: thanks! | 12:54 |
StevenK | lifeless: TBH, I came to Dublin not expecting to do anything like Skype. I can only do Mumble, except that I don't have a headset. | 12:54 |
lifeless | StevenK: the size of the data doesn't particular impact the service-nature of it for me | 12:55 |
lifeless | StevenK: things that make a difference are - is it potentially reusable for other goals? | 12:56 |
StevenK | lifeless: It is not. | 12:56 |
lifeless | StevenK: does it have different HA needs to the rest of the system (e.g. if its a cache we could rebuild it on failure rather than having N copies) | 12:56 |
StevenK | It's a list of contents of binary packages, I can't think of anything aside from contents generation that will find that useful. | 12:56 |
wgrant | StevenK: But, equally, there's no reason to have it in LP :) | 12:57 |
StevenK | But it's *easy* if it is in LP. | 12:57 |
lifeless | StevenK: does it have different IO needs than the rest of the system - e.g. will having it in the main DB sacrifice memory to it whereas on a different system it could be always-cold and we wouldn't care | 12:57 |
StevenK | And since this is a spare-time project, I'm loving the easy. | 12:57 |
wgrant | StevenK: But it's in LP if it is in LP. | 12:57 |
lifeless | does the rest of the system need to know about the implementation of this thing? | 12:58 |
StevenK | The publisher will | 12:58 |
lifeless | (if yes, it shouldn't be a service; if no it could be a service) | 12:58 |
lifeless | StevenK: FWIW I think that the contents.gz thing wouldn't make a good microservice on its own; it would need publishing data included in it. Such redundancy isn't necessarily bad (in fact having a service that handles archives, their internal consistency, etc, including contents.gz, would be awesome) | 13:06 |
lifeless | StevenK: so I'm totally fine with you including it in LP itself, given that we don't have that other service to put this in instead. | 13:06 |
StevenK | lifeless: Okay. | 13:06 |
lifeless | StevenK: it will often be easier (short term) to bundle stuff into the LP tree | 13:08 |
lifeless | StevenK: I'm glad stub suggested making it into a microservice; I think this particular case is probably too awkward right now, but we need to start doing it more | 13:09 |
lifeless | alrighty, its pumpkin time | 13:12 |
lifeless | ciao | 13:12 |
bigjools | matsubara: hey, sorry didn't answer you earlier | 14:05 |
=== jr__ is now known as Riddell | ||
jtv | wgrant, StevenK: do you have uncommitted changes in dogfood's source tree? | 14:49 |
StevenK | Yes. | 14:50 |
jtv | boo | 14:50 |
nhandler | I'm writing a Perl script to try and interact with Launchpad using the API. I am able to get a request and access token, but making signed API requests fails. I get a 401 error with a 'Client-Warning: Missing Authenticate header' (despite having a request matching that of the documentation) and an 'Unknown consumer (nameOfMyApp)' message. Any ideas on what might be wrong? | 15:19 |
jcsackett | wallyworld: https://code.launchpad.net/~launchpad-teal-squad/launchpad/lazr-js-kicking-and-screaming | 15:23 |
rvba | wgrant: could you send me these cryptic python lines that you used to publish a message on rabbit? | 15:28 |
wgrant | rvba, bigjools: http://paste.ubuntu.com/634325/ | 15:29 |
huwshimi | benji: https://code.launchpad.net/~huwshimi/launchpad/what-next-712259/+merge/66030 and https://code.launchpad.net/~huwshimi/launchpad/table-headings-728187/+merge/66008 | 15:31 |
wgrant | bigjools, rvba: http://paste.ubuntu.com/634326/ | 15:31 |
* benji looks. | 15:31 | |
jcsackett | huwshimi: https://code.launchpad.net/~launchpad/launchpad/lazr-js-kicking-and-screaming | 15:34 |
wallyworld_ | rockstar: ping! | 15:37 |
benji | huwshimi: both branches are approved, I had a thought on the second you might be interested in (or not;) | 15:48 |
huwshimi | benji: Thanks a lot, I'll take a look | 15:48 |
bigjools | stub: http://paste.ubuntu.com/634326/ | 16:00 |
rockstar | wallyworld_, pong | 16:04 |
bigjools | rvba: http://pastebin.ubuntu.com/634349/ | 16:04 |
wallyworld_ | hi rockstar. long story. we are packaging yui as a tarball to go in download-cache and i am porting the lazr-js build utils across to lp. someone said you had done something similar. i was wondering about any gotchas that i should be aware of | 16:06 |
rockstar | wallyworld_, https://launchpad.net/lazr.jstools | 16:08 |
wallyworld_ | rockstar: thanks. i'll take a peek | 16:08 |
rockstar | wallyworld_, hm, there seems to be no branch there, but I'm sure I pushed it. | 16:08 |
rockstar | wallyworld_, lemme dig around, I'm sure I have a backup of it as well. | 16:09 |
wallyworld_ | rockstar: ok. thanks. appreciate it | 16:09 |
rockstar | wallyworld_, I will say that the tools we were using in lazr-js tend to be less effective than, say YUI Compressor. | 16:09 |
rockstar | U1 is now using YUI compressor. It compresses YUI code a lot better than Chrome's js tools we were using in lazr-js. | 16:10 |
bigjools | rvba: http://pastebin.ubuntu.com/634351/ | 16:10 |
wallyworld_ | rockstar: ok. we'll need to take a look at that too then. first step is a straight port and then we can tweak and optimise :-) | 16:10 |
danilos | jtv, https://code.launchpad.net/~danilo/launchpad/branch-portlet-details-remove/+merge/66117 | 16:18 |
danilos | poolie, I've added the link to jtv's branch to the project wiki page, fwiw | 16:25 |
poolie | thanks! | 16:29 |
wgrant | rvba: http://paste.ubuntu.com/634369/ | 16:41 |
rvba | wgrant: thanks. | 16:41 |
jml | Narwhals! Narwhals! | 16:45 |
wgrant | Swimming in the ocean. | 16:46 |
poolie | Riddell: https://bugs.launchpad.net/bugs/801388 | 16:46 |
_mup_ | Bug #801388: some person pickers show "assign me"/"remove assignee" when that makes no sense <disclosure> <person-picker> <ui> <Launchpad itself:In Progress by wallyworld> < https://launchpad.net/bugs/801388 > | 16:46 |
jml | jcsackett: http://metaquotes.livejournal.com/6644038.html | 16:49 |
jelmer | jml: Do you perhaps have a moment to go over the BFBIA LEP this week? | 16:56 |
jml | jelmer: yes, I do. | 16:56 |
jml | jelmer: could do it pretty soon, actually | 16:57 |
jml | jelmer: otherwise, make an appointment on my calendar. | 16:57 |
jelmer | jml: Pretty soon works for me, too | 16:58 |
jml | jelmer: ok. | 16:58 |
LPCIBot | Project parallel-test build #79: STILL FAILING in 1 hr 1 min: https://lpci.wedontsleep.org/job/parallel-test/79/ | 17:00 |
LPCIBot | Yippie, build fixed! | 17:05 |
LPCIBot | Project devel build #844: FIXED in 5 hr 52 min: https://lpci.wedontsleep.org/job/devel/844/ | 17:05 |
=== beuno is now known as bueno-lunch | ||
=== salgado is now known as salgado-lunch | ||
danilos | deryck, hey, since sinzui is not on IRC, how about you look at https://code.launchpad.net/~danilo/launchpad/speed-up-subscription-tests/+merge/66120 :) | 17:07 |
deryck | danilos, sure. :-) Looking now.... | 17:08 |
deryck | danilos, Y.lazr.anim was already available due to YUI().use for the test module? | 17:10 |
danilos | deryck, yeah, and fwiw, the test works | 17:10 |
danilos | deryck, I also increased the wait to 200ms from 20ms on sinzui's suggestion | 17:10 |
deryck | danilos, I'm worried the 200ms is still too long. But there can be some difference between machines.... | 17:11 |
deryck | danilos, but I would think 100ms should be enough even for slower machines. | 17:11 |
deryck | unless the dom work is happening beyond the anim completing. | 17:12 |
deryck | danilos, but sinzui tells me know 100-200ms is expected. So I say try 100. Lower is always better :) | 17:13 |
danilos | deryck, 20ms works for me just fine, but let's try 100ms for buildbot (if it ever gets back there :) | 17:23 |
deryck | danilos, I'm chatting more with sinzui about this now. his machine is what failed earlier, not buildbot. | 17:23 |
danilos | deryck, oh | 17:23 |
=== bueno-lunch is now known as beuno | ||
LPCIBot | Yippie, build fixed! | 18:20 |
LPCIBot | Project db-devel build #675: FIXED in 5 hr 53 min: https://lpci.wedontsleep.org/job/db-devel/675/ | 18:20 |
=== salgado-lunch is now known as salgado | ||
lifeless | flacoste: hi | 22:08 |
LPCIBot | Project devel build #845: FAILURE in 5 hr 34 min: https://lpci.wedontsleep.org/job/devel/845/ | 22:39 |
poolie | lifeless, Riddell, maybe we should just remove that option | 23:07 |
poolie | does anyone ever want to see the least recently changed bugs? | 23:07 |
poolie | i suppose maybe if you want to update the stalest inprogress or incomplete bugs | 23:08 |
* jelmer doesn't think he's ever used it or would ever use it | 23:08 | |
poolie | perhaps 'changed long ago' | 23:09 |
poolie | kind of gets into whether people ven know what 'changed' means, exactly | 23:09 |
poolie | the bug in this area i would actuall ymost like closed is https://bugs.launchpad.net/launchpad/+bug/277352 | 23:11 |
_mup_ | Bug #277352: should be easier to search for closed bugs <lp-bugs> <Launchpad itself:Triaged> < https://launchpad.net/bugs/277352 > | 23:11 |
poolie | (which is not so trivial, but maybe not so hard) | 23:11 |
poolie | i'm surprised it doesn't have more dupes | 23:11 |
poolie | lifeless: when's the next lpnet rollout? | 23:15 |
jelmer | poolie, that one has annoyed me too in the past | 23:19 |
lifeless | poolie: when someone in dubling requests it | 23:21 |
poolie | is it cheap? can i just request it? | 23:24 |
lifeless | if you follow the process, yes. | 23:27 |
lifeless | one bit of which is 'do not deploy right before everyone goes to sleep | 23:28 |
lifeless | normally thats friday afternoons, but during sprints, any deploy except first thing in the morning would qualify ;) | 23:28 |
wgrant | poolie: There's also nothing to deploy right now. | 23:29 |
wgrant | Lots of QA pending. | 23:29 |
wgrant | rvb/abentley/jtv/matsubara | 23:29 |
poolie | :) | 23:30 |
poolie | ah this'd be https://devpad.canonical.com/~lpqateam/qa_reports/deployment-stable.html ? | 23:30 |
wgrant | Yes | 23:31 |
poolie | wow, that's quite a queue | 23:31 |
poolie | thanks | 23:31 |
lifeless | we can deploy any qa | 23:31 |
lifeless | 'd prefix of it | 23:31 |
lifeless | wgrant: courier rang, new machine arrives in the next 4 hours | 23:32 |
wgrant | lifeless: ! | 23:32 |
wgrant | poolie: We deployed 30 revs yesterday. | 23:32 |
poolie | nice | 23:32 |
wgrant | No. | 23:33 |
wgrant | Because the queue was huge. | 23:33 |
wgrant | And still is. | 23:33 |
poolie | lifeless: what model was it? | 23:33 |
poolie | s//is | 23:33 |
lifeless | poolie: dell aurora r3 4(8) core w/16GB | 23:34 |
lifeless | the cpu and mb should handle 32gb but you cannot source the dimms at the moment | 23:38 |
poolie | wow, overclocked from the factory? | 23:40 |
lifeless | wgrant: so I know its not finished, but I need your aufs notes ;) | 23:48 |
wgrant | lifeless: Let me see if WOL works. | 23:50 |
wgrant | It does. | 23:50 |
lifeless | wgrant: + your testr config etc etc. :> | 23:50 |
wgrant | http://paste.ubuntu.com/634628/ | 23:51 |
wgrant | http://paste.ubuntu.com/634629/ | 23:53 |
wgrant | Then 'TEMP=/tmp/testr testr run --parallel' | 23:54 |
wgrant | (yes, the /tmp/testr hack is terrible, but meh, it works) | 23:54 |
lifeless | I'll deal | 23:54 |
wgrant | This expects the lucid-lp-base container to have had rocketfuel-setup, utilities/launchpad-database-setup and make schema run. | 23:55 |
wgrant | I'm in no fit state to experiment further now, but if you have any issues I can try to recall now or exeriment tomorow. | 23:57 |
lifeless | well | 23:57 |
lifeless | I'm going to go make sure I have a natty boot cd | 23:57 |
lifeless | I expect backing up the windows install and fiddling around data migration etc will take a chunk of time today | 23:57 |
wgrant | Yeah. | 23:57 |
wgrant | You forgot the ritual zeroing of the disk to ensure obliteration of the Windos installation. | 23:58 |
wgrant | Most of the problems I encountered were around dbus/avahi issues, which are irrelevant now I extract the last DHCP lease. | 23:59 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!