The web in a box - a next generation web framework for the Perl programming language

IRC log for #mojo, 2017-06-26

| Channels | #mojo index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:09 cheako joined #mojo
00:55 cheako joined #mojo
01:05 zivester joined #mojo
01:15 schelcj joined #mojo
01:20 howitdo joined #mojo
01:45 bb9 joined #mojo
02:02 lthemick joined #mojo
02:14 schelcj joined #mojo
02:14 karjala_ joined #mojo
02:19 noganex joined #mojo
02:33 bb9 joined #mojo
02:44 schelcj joined #mojo
03:37 schelcj joined #mojo
04:05 new_student joined #mojo
04:08 new_student Hi! Could someone help me debug why mojo is using 100% of my CPU? http://picpaste.com/pics/Screenshot_from_2017-06-26_09-36-08-In2eLnCF.1498450058.png
04:09 new_student It happens only when I bombard the server with too many requests
04:09 new_student I am using Mojolicious:Lite with the ForkCall plugin
04:12 new_student For now, I just get around this by adding a cron job to restart the api server every night :'(
04:14 Grinnz not sure i like this query https://github.com/Grinnz/Minion-Backend-SQLite/blob/master/lib/Minion/Backend/SQLite.pm#L108 if someone tries to set a key with a . or [] in it they're going to get a weird structure back
04:15 Grinnz but i don't know if sqlite json paths can be escaped
04:19 Grinnz i could just throw an error if the user tries to set a key containing those characters.. heh
04:32 inokenty-w joined #mojo
04:36 cheako joined #mojo
04:57 Grinnz 00:47:14 — GumbyPAN CPAN Upload: Minion-Backend-SQLite-2.000 by DBOOK https://metacpan.org/release/DBOOK/Minion-Backend-SQLite-2.000
05:05 ssm joined #mojo
05:29 brunoramos joined #mojo
05:29 karjala_ joined #mojo
05:45 cheako joined #mojo
05:49 cheako left #mojo
06:08 ashimema joined #mojo
06:23 dod joined #mojo
06:26 disputin joined #mojo
06:28 dod joined #mojo
06:45 AndrewIsh joined #mojo
07:00 karjala_ joined #mojo
07:22 prg joined #mojo
07:22 trone joined #mojo
07:32 sri Grinnz++ # that was fast
07:44 onix joined #mojo
08:01 dod joined #mojo
08:13 stryx` joined #mojo
08:18 rshadow joined #mojo
08:50 Lee joined #mojo
08:59 rshadow joined #mojo
09:05 cheako joined #mojo
09:18 CandyAngel Hm, can I output a stringified DOM element without it showing its children? I'm only interested in the element and its attributes..
09:20 rshadow joined #mojo
09:34 cheako joined #mojo
10:06 prajith joined #mojo
10:06 prajith any idea about this Can't locate object method "get_handle" via package "SysOps::I18N" ?
10:07 prajith I have already added this "use parent 'Locale::Maketext';" to 'SysOps::I18N' package
10:18 prajith fixed it thanks :)
10:21 cheako joined #mojo
15:07 ilbot2 joined #mojo
15:07 Topic for #mojo is now ???? nom nom | http://mojolicious.org | http://irclog.mojolicious.org | http://code-of-conduct.mojolicious.org
15:10 trwww1 joined #mojo
15:12 disputin joined #mojo
15:14 disputin joined #mojo
15:27 kamyl joined #mojo
16:14 lthemick1 joined #mojo
16:16 howitdo joined #mojo
16:44 cheako joined #mojo
16:51 jabberwok It looks like Pg is the only current backend for Minion so far... I am guessing Sqlite doesn't have the guts for it?
16:52 Grinnz what? there are several https://metacpan.org/search?q=minion+backend
16:52 Grinnz i wouldn't use the sqlite backend in a high load scenario, though
16:52 jabberwok oh. right.  i knew that. (( apply directly to forehead ))
17:10 sri only Pg and SQLite backends are up to date though
17:10 sh14 joined #mojo
17:11 sri and i guess Storable... but that's not really useful ;p
17:12 jabberwok that's perfect actually.
17:12 dod joined #mojo
17:20 rshadow joined #mojo
17:51 PryMar56 joined #mojo
18:14 cheako joined #mojo
18:21 itaipu joined #mojo
18:33 tchaves joined #mojo
19:09 cheako left #mojo
19:10 bb9 joined #mojo
19:34 karjala_ joined #mojo
19:45 karjala_ Do people here know whether LXD containers are production-ready? Would you trust an important site to contain it in an LXD container?
19:48 preaction isn't that what docker's been doing since it was created?
19:49 karjala_ I couldn't get docker to work well for me
19:49 karjala_ LXD might be different in that containers are complete OSes
19:49 karjala_ Although I think the two can work together
19:49 karjala_ docker & LXD
19:50 karjala_ I think I read that
19:50 preaction so... it's like docker, but pid 1 is init(1)
19:50 karjala_ And very easy to use (which docker isn't)
19:50 preaction i mean, my opinion would be it's built on the same features as docker, and so i'd trust it approximately as much
19:51 preaction there was a talk at TPC about docker, and it seemed really super easy to me
19:51 karjala_ oh ok
19:53 sri fwiw. i've yet to see someone use lxd
20:00 sri oh, it's not even available in opensuse ;p
20:00 sri choice of operating systems seems pretty limited
20:01 preaction yeah, that's mostly why i don't use all that stuff: none of it works on BSDs :p
20:05 karjala_ lxd exists on snappy
20:05 karjala_ as well
20:05 karjala_ so maybe if suse has snap, it can (maybe) run lxd
20:05 karjala_ maybe not
20:06 karjala_ :-)
20:12 nic I'm using Minion::Backend::Sereal at work, but the throughput of jobs is very low -- jobs take around 20 mins, so it's a mean rate of 12 jobs/hour :)
20:14 sri nic: i kinda consider all backends that can't run the benchmark just toy projects
20:15 nic which ones can't run the benchmark?
20:15 sri storable can't
20:16 nic hmm, ok, I accept I may have done something wonky, but it appears to run the benchmark fine
20:16 sri and the DBM::Deep one couldn't either
20:17 nic do you mean with high performance?  It's true that ::Storable maxes out at around 15 jobs/sec on my laptop
20:17 nic and ::Sereal maxes out at around 40 jobs/sec
20:17 sri what are your results on https://st.aticpan.org/source/SRI/Minion-7.01/examples/minion_bench.pl
20:18 nic that's the bm I'm using
20:18 nic unfortunately I deleted all my results this afternoon cos I'd convinced myself there was no point tuning further
20:18 sri odd, i remember it not working, although i gave up on storable a long time ago
20:19 sri i do remember DBM::Deep came to a complete halt with concurrent workers
20:20 nic I spent some time at the weekend doing tuning, because it _appeared_ that I was getting macro results, but then I realised I was micro-optimising -- because concurrency is poor on these backends, a tiny performance improvement looks significant because you're actually reducing collisions
20:20 sri i just can't imagine rewriting the storable file for 10k jobs every time you dequeue, and then again when you finish the job could work for 4 parallel workers
20:21 sri even if you accept that you're killing your ssd
20:21 nic yeah if you have either (a) 20k jobs, or (b) 8+ workers, you don't want one of my NoneSQL platforms :)
20:22 sri so, you modified the benchmark?
20:22 nic I ran the bm with 2 loops and 2 workers
20:23 nic actually, to check my code, I ran with just 1 worker, to avoid the concurrency effect, then went to 2 workers to check that side too
20:23 sri i see very little reason not to just use SQLite
20:23 sri which actually performs pretty well on the benchmark
20:24 nic I think the fs-based, I mean NoneSQL(TM), platforms are really only for (a) learning & playing, (b) prototyping, (c) embedded platforms
20:25 sri you're still killing your ssd
20:26 nic that depends the rate of changes you're making
20:27 nic a lot of background-processing situations are pretty low-throughput
20:27 sri i've actually not encountered those yet
20:28 sri mine end up being thousands of jobs per day
20:28 nic eg the deployment of ::Sereal, the web app needs to kick off a 20min SQL report and then email the user that it's ready (as an Excel download)
20:28 jberger nic: even in those cases, why not use SQLite?
20:28 nic fewer moving parts
20:29 nic don't get me wrong, I value the SQLite backend, and if I didn't have the Roach backend I'd probably be using it for my next project
20:30 Grinnz the SQLite version of the benchmark is slightly modified too, but only in the "setup" part because sqlite is weird with forks
20:31 Grinnz also, doing an analyze on the table made it lock up for some reason
20:31 Grinnz but without that, it still works :P
20:32 nic https://github.com/niczero/minion-backend-storable/blob/master/test/minion_bench.pl#L9 is the only tweak I made
20:32 nic sorry that file doesn't credit who wrote it -- I'll fix that tonight
20:33 nic (although the entire package does credit the author as a co-author :) )
20:46 Pyritic joined #mojo
21:11 bb9 joined #mojo
22:12 Peppard joined #mojo
22:13 rshadow joined #mojo
23:29 gizmomathboy joined #mojo

| Channels | #mojo index | Today | | Search | Google Search | Plain-Text | summary