Perl 6 - the future is here, just unevenly distributed

IRC log for #metacpan, 2016-07-06

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

All times shown according to UTC.

Time Nick Message
00:34 jdv79 depends:)  ideally never but caching is hard.
00:35 jdv79 not sure wha you hit since i cant repro
00:36 jdv79 sounds awfully like http caching though without knowing more.
01:59 haarg we have plans that will fix that
02:00 haarg it isn't an especially hard problem to solve, but we don't have much manpower and it has recently been invested elsewhere
02:36 punter nice
03:02 metacpan joined #metacpan
03:02 metacpan [metacpan-api] oalders pushed 1 new commit to master: https://git.io/vKICt
03:02 metacpan metacpan-api/master 1f1f048 Olaf Alders: Merge pull request #501 from metacpan/mickey/test_backpan...
03:02 metacpan left #metacpan
03:02 metacpan joined #metacpan
03:02 metacpan [metacpan-api] oalders deleted mickey/test_backpan at 1175923: https://git.io/vKICq
03:02 metacpan left #metacpan
03:55 jdv79 haarg: what is the problem succintly?
03:58 haarg we aren't serving appropriate caching headers
03:58 jdv79 :(
03:58 jdv79 tisk tisk:)
03:58 haarg the solution will be to serve caching headers so that our cdn caches indefinitely, browsers must-revalidate, and then we invalidate the cdn caches as we index things
03:59 jdv79 good luck with that.  caching is hard.
04:00 jdv79 been there don't envy being there
04:00 haarg it won't be too hard, it just touches enough stuff that it isn't trivial either
04:00 jdv79 likely the tip of the iceberg
04:00 jdv79 let me know when that's all sorted:)
04:01 jdv79 haha
04:01 haarg we've already done the preliminary work - there is nothing specific to the user on any of the pages outside of the account stuff.
04:02 jdv79 cool
04:17 vanstyn joined #metacpan
05:22 ranguard oalders: please point people to us: fastapi.mc.org
05:22 ranguard not api-v1.mc.org
05:22 ranguard (re blog post about download_url)
06:35 oiami joined #metacpan
07:01 mickey also to MetaCPAN::Client :)
07:02 mickey though it is true that MetaCPAN::API doesn't support the new endpoint :)
07:04 [Tux] <h1>Error 503 All backends failed or unhealthy</h1>
07:04 [Tux] <p>All backends failed or unhealthy</p>
07:04 [Tux] <h3>Guru Mediation:</h3>
07:04 [Tux] <p>Details: cache-ams4150-AMS 1467779443 344102087</p>
07:04 [Tux] Date: Wed,  6 Jul 2016 06:30:43 +0200 (CEST)
07:05 metacpan joined #metacpan
07:05 metacpan [metacpan-client] mickeyn pushed 2 new commits to master: https://git.io/vKI23
07:05 metacpan metacpan-client/master 063d150 Mickey Nasriachi: Set default 'domain' value if 'version' is v0|v1
07:05 metacpan metacpan-client/master 929c549 Mickey Nasriachi: Support download_url endpoint
07:05 metacpan left #metacpan
07:10 metacpan joined #metacpan
07:10 metacpan [metacpan-client] mickeyn pushed 1 new commit to master: https://git.io/vKI28
07:10 metacpan metacpan-client/master 381c727 Mickey Nasriachi: release 1.018000
07:10 metacpan left #metacpan
07:10 mickey oalders: this new version of MetaCPAN::Client supports the download_url ;)
07:19 neilb joined #metacpan
08:10 metacpan joined #metacpan
08:10 metacpan [metacpan-client] mickeyn pushed 2 new commits to master: https://git.io/vKIoB
08:10 metacpan metacpan-client/master 8bf7f88 Mickey Nasriachi: fix abstract
08:10 metacpan metacpan-client/master 5ba9293 Mickey Nasriachi: added test api/download_url
08:10 metacpan left #metacpan
08:11 neilb joined #metacpan
08:15 punter joined #metacpan
08:17 neilb joined #metacpan
09:24 Relequestual joined #metacpan
13:11 punter joined #metacpan
13:42 oalders ok ok. i'm going to have people start proofreading my work :)
13:42 oalders mickey++
13:42 oalders will update appropriately
13:44 oalders [Tux] was this just a one time error?
13:59 metacpan joined #metacpan
13:59 metacpan [metacpan-client] mickeyn pushed 1 new commit to master: https://git.io/vKLcq
13:59 metacpan metacpan-client/master ff77d7d Mickey Nasriachi: Added missing 'download_url' key in 'file' type
13:59 metacpan left #metacpan
14:05 metacpan joined #metacpan
14:05 metacpan [metacpan-client] mickeyn pushed 1 new commit to master: https://git.io/vKLCu
14:05 metacpan metacpan-client/master ec8057c Mickey Nasriachi: release: 1.019000
14:05 metacpan left #metacpan
14:48 oalders mickey,ranguard: blog post updated
14:50 mickey oalders++ # spreading the word
14:51 ranguard ++
15:26 Khisanth joined #metacpan
15:39 Tux oalders, yes, but I get on average 10 mails a day, and I run the dashboard generator every 15 minutes
15:39 Tux 13,28,43,58 * * * * /work/www/make-perl.pl
16:36 neilb joined #metacpan
17:06 alh_ joined #metacpan
18:35 oiami joined #metacpan
18:53 [Tux] Date: Wed,  6 Jul 2016 20:37:36 +0200 (CEST)
18:53 [Tux] [Timeout] ** [https://fastapi.metacpan.org:443]-[599] Timed out while waiting for socket to become ready for reading, called from sub Search::Elasticsearch::Role::Client::Direct::Main::scroll_helper at /pro/lib/perl5/site_perl/5.22.0/MetaCPAN/Client/Request.pm line 108. With vars: {'status_code' => 599,'request' => {'mime_type' => 'application/json','method' => 'POST','path' => '/v1/favorite/_search','ignore' => [],'body' => {'query' => {'term' =>
18:53 [Tux] {'distribution' => 'Tk-Clock'}}},'serialize' => 'std','qs' => {'size' => 1000,'scroll' => '5m','search_type' => 'scan'}}}
18:55 oalders ranguard: do we have any idea of how often these timeouts are happening?
19:53 ranguard oalders: looking
19:58 neilb_ joined #metacpan
20:02 ranguard oalders: created https://github.com/metacpan/metacpan-api/issues/503 ( heh, good issue number!) - but that's for general 500 errors
20:02 ranguard oh actually that does seem to relate to the scroller
20:04 ranguard https://www.evernote.com/shard/s54/sh/f54d9b2c-4c91-425a-bc78-e1f0ab6fb47f/d555311e37389cfacfdbf73a939516b0 are the actual timeouts (where fastly gave up, 15s for connection and 50s for initial response)
20:08 oalders ranguard: thanks. can we tweak the timeout limits?
20:09 oalders are we logging slow queries?
20:13 ranguard oalders: we can tweak already (went live 22nd June) - but yea we can tweak further if needed
20:14 oalders so those numbers are just arbitrary?
20:16 ranguard yea, it's what we have on metacpan.org I think - but all tweakable
20:17 ranguard but does look like there's an issue with scrolling deletes.. actually should be just be blocking those at fastly?
20:18 ranguard 'DELETE' methods requests I mean?
20:20 oalders well, with the dogfooding, that's more complicated
20:20 oalders there's a legitimate case for a DELETE from the from the front end (for removing a ++)
20:21 oalders do we have an idea of how using fastly on the front end is affecting response times?
20:22 ranguard api should connect directly, just mc-web should eat the dog food :)
20:22 ranguard why would mc-web need to DELETE ?
20:23 neilb_ joined #metacpan
20:24 oalders ranguard: is that not what happens when we remove a ++? (haven't checked)
20:24 * ranguard shrugs :)
20:25 oalders there's a DELETE when a user removes an identity
20:25 oalders there's a DELETE for removing a ++
20:26 oalders lib/MetaCPAN/Web/Model/API/User.pm
20:26 ranguard yea, just got there
20:26 * ranguard could block DELETE for all but our IP's
20:27 oalders that would be good
20:28 ranguard it's 80.100.130.165 that's calling the DELETE
20:29 ranguard [Tux]: that you?
20:29 [Tux] I think so
20:29 oalders heh
20:29 [Tux] do I call delete on anything?
20:30 [Tux] and yes, that is me
20:30 ranguard https://gist.github.com/ranguard/9edfe206c6e54964b451b016e4e36884 some of your requests recently
20:31 * [Tux] wonders how
20:31 [Tux] I really just want read-only inquiring
20:32 * ranguard will be blocking shortly in anycase :)
20:32 [Tux] and the cron page creator is just using http, no MetaCPAN::Client yet
20:34 [Tux] have I been hacked?
20:34 * ranguard would expect a bug first :)
20:37 [Tux] my $data = eval { $mcpan->module ($mod)->{data} } || {};
20:37 [Tux] if ($m->{data}) {
20:37 [Tux] $data->{$_} = $m->{data}{$_} for keys %{$m->{data}};
20:37 [Tux] }
20:37 [Tux] that might be the cause
20:37 [Tux] Oke, I need a READONLY flag
20:39 [Tux] bedtime
20:46 mickey i think the DELETEs are scroller cleanup issued by Search::Elasticsearch
20:47 mickey https://www.elastic.co/guide/en/elasticsearch/reference/current/search-request-scroll.html#_clear_scroll_api
20:50 mickey blocking that will be bad
20:51 oalders yep
20:52 ranguard mickey: well they are all throwing 500 errors atm
20:54 mickey anything in the logs?
20:55 mickey i'm far from a computer atm to check myself :)
21:03 * mickey says nn
21:08 Grinnz_ v1.metacpan.org isn't showing any abstracts on the release pages, is this a known issue?
21:10 oalders Grinnz_ : example?
21:11 Grinnz_ https://v1.metacpan.org/release/JV/App-Music-ChordPro-0.63
21:11 Grinnz_ compare with https://metacpan.org/release/JV/App-Music-ChordPro-0.63
21:11 oalders ah, thanks
21:11 oalders Grinnz_ : could you open an issue on metacpan-web?
21:11 Grinnz_ sure
21:40 * ranguard nn's as well
22:03 oalders Grinnz_ : thanks
22:06 _dolmen_ joined #metacpan
22:26 punter joined #metacpan
22:47 punter joined #metacpan
22:50 punter joined #metacpan
23:17 neilb_ joined #metacpan

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