Perl 6 - the future is here, just unevenly distributed

IRC log for #metacpan, 2016-07-15

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

All times shown according to UTC.

Time Nick Message
05:09 metacpan joined #metacpan
05:09 metacpan [metacpan-web] mickeyn created mickey/GH1692 (+1 new commit): https://git.io/vKEJ1
05:09 metacpan metacpan-web/mickey/GH1692 f55c33d Mickey Nasriachi: author page: cpan/backpan counts + backpan-only message
05:09 metacpan left #metacpan
06:36 Khisanth joined #metacpan
07:39 oiami joined #metacpan
08:03 Relequestual joined #metacpan
08:11 neilb joined #metacpan
08:36 neilb joined #metacpan
11:51 castaway joined #metacpan
12:31 punter joined #metacpan
13:42 cjm joined #metacpan
14:41 mickey oalders: re GH issues: guessing this is a 'Pithub + GH token + v1' issue (errors: '{"message":"API rate limit exceeded for 50.28.18.168. (But here\'s the good ne
14:41 mickey ws: Authenticated requests get a higher rate limit. Check out the documentation for more details.)","documentation_url":"https://deve
14:41 mickey loper.github.com/v3/#rate-limiting"}'
14:41 mickey sorry for the broken paste :)
14:42 oalders :)
14:42 oalders so we're not using a token for the api calls?
14:42 mickey apparently, that's the error in the tickets script
14:43 oalders ok. that's easy enough to fix.  thanks for finding that
14:44 mickey cheers, please let me know if you need me to dig into it more / test it
14:48 oalders thanks. i'll get it using the token and also look into the github app for v1 while i'm at it
15:11 jnap joined #metacpan
15:12 jnap anyone know if there's a pending issue?  I can't spot anything wrong with my module metadata and it seems fine on search.cpan.org
15:12 jnap https://metacpan.org/pod/Template::Pure  FWIW, points to the March 2016 release and the release from this week is marked on backspin :)
16:02 Grinnz_ seems to work correctly on v1, maybe the v0 index is out of whack
16:02 Grinnz_ https://v1.metacpan.org/pod/Template::Pure
16:04 jnbek joined #metacpan
16:12 neilb joined #metacpan
16:24 oalders jnap: it's back now.  sorry about that. had an issue earlier this week
16:38 jnap oalders: no worries at all, just thought to ask.  thanks.  oalders++
16:39 neilb joined #metacpan
17:07 punter joined #metacpan
18:13 ranguard puppet now runs one the hour, every hour on the lw-* boxes
18:13 ranguard last run should be logged to: /var/log/puppet/last_run.log
18:14 * ranguard will check when the hour rolls around
18:16 ranguard mickey: you mentioned lw-3 memory
18:16 ranguard http://munin.metacpan.org/munin-cgi/munin-cgi-html/metacpan.org/lw-mc-03.metacpan.org/memory.html - the 'cache' in blue is just that, OS caching stuff, but that would be dropped for application usage
18:17 ranguard so we've still got about 8G free most of the time
18:17 ranguard use `free -m` - and look at the `-/+ buffers/cache` line
18:18 ranguard though I don't like that stuff is in swap
18:19 ranguard mst: remind me why I shouldn't care again please?
18:22 kentnl nb: "swap_cache" is "stuff that is both in memory and in swap, so it can access it immeditely, but it can dump it immediately if there's not much ram free for something else more important"
18:23 kentnl and cache is pretty much considerable to be "Free memory, just files are in there because you want IO to not go directly to disk all the time because that's slow"
18:23 kentnl yeah.
18:24 kentnl so swap is basically "we got rid of stuff your app wasn't using and will probably never use because IO is more importanter"
18:25 ranguard kentnl: on a `free -m` that's showing swap, not swap_cache right?
18:26 ranguard Swap looks to be the big red band in munin graph
18:28 ranguard at work I only see swap when an app goes nuts and uses too much memory, otherwise it's 0
18:29 ranguard so I've just got Swap==bad/IO issue in my head
18:29 kentnl Yeah.  The question is  really though "Are we reducing IO by using Swap or not?"
18:30 kentnl if you're going to do 100,000 iops on a block of disk, its better to do that in ram than do it directly. Esp if there's a block of memory that hasn't been touched in hours
18:31 nicomen I thought people just turned off swap on servers
18:32 nicomen better to get an OOM and get it overwith. If it's a leak you set up a loop that restarts the service, if it's too few resources, you fix code/hardware.
18:33 kentnl its useful to have swap in some cases.
18:33 nicomen Also this talk seemed to talk a bit about memory/swap on linux: https://www.youtube.com/watch?v=QG9FIFKDKOU
18:33 nicomen (from yapc na)
18:34 kentnl just see "The more ram you have, the more of that ram can be used to turn rust into faster-than-SSD" ;)
18:44 avenj I like having plenty of swap. default linux vm behavior is pretty reasonable with regards to swapping out inactives to make room for fs cache, and also tunable
18:55 ranguard thanks all
19:22 neilb joined #metacpan
19:24 nicomen when I ran out of memory, it seldomly recovers, always end up swapping forever, which makes machines inaccessible
19:24 nicomen *run
19:56 ribasushi Swap:     39845884   17027624   22818260
19:56 ribasushi ^^ my workstation experience would be unthinkable without swap
19:58 kentnl there are problems in Linux's OOM handling that are getting slowly fixed. But yeah, if you're getting swap-rushed, you're likely to be tempting well outside OOM conditions anyway.
19:58 kentnl But it all of course depends on workloads.
19:59 kentnl its like "which is worse, waiting 30 seconds for some part of firefox to come back from swap, or waiting half an hour for `git status` on a large repo?"
20:00 * kentnl numbers of course not really representative
20:25 nicomen I thought we were talking servers, for workstations I use swap
20:25 nicomen as a reboot when everything is responsive is acceptable
20:25 nicomen *unresponsive
20:26 nicomen but the talk I pasted was interesting, the parts about swap is something I want to check out more thoroughly. I think Linux has some non-optimal defaults for swap, similar to issues with scheduling
23:13 neilb_ joined #metacpan

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