19:32:21 #startmeeting 19:32:21 Meeting started Wed Mar 25 19:32:21 2015 UTC. The chair is TheSnide. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:32:21 Useful Commands: #action #agreed #help #info #idea #link #topic. 19:33:21 hi all 19:33:51 'ello 19:33:56 Hi! 19:33:57 #info 2.1.11 is out. 19:34:09 and there was much rejoicing 19:34:14 Yaaay 19:34:42 :D 19:34:56 Any highlights for the 2.1.11 release? 19:35:00 Do you know approximatively when it will be available on main distros? 19:35:15 munin-cgi-* is dead, long live munin-httpd? 19:36:24 chteuchteu: I've spent some time on the packaging. It'll take a bit more time to get all the packaging metadata (Depends, Breaks, Conflicts, Build-Depends) right, as there also will be more munin-plugins-* packages. Smooth upgrades from previous versions must also work 19:37:16 Separate packages for munin-plugins-{core,snmp,postgresql,mysql,whatever,ilikeminions,extras} 19:39:07 #info 2.1.11 contains munin-httpd that *replaces* allcgi. 19:39:28 . o O { and there was much rejoicing } 19:40:01 I think the cgi scripts have been the primary source of confused users on this channel :) 19:40:10 as it was based on the field feedback that cgi was nice to have, but hard to configure 19:40:18 indeed 19:41:07 now it just takes going to http://localhost:4948/ to enjoy everything 19:41:39 #info munin will not (for the forseeable future) have ssl nor acl in httpd 19:42:08 what about http/2? ^.^ 19:44:48 * ssm was not entirely serious about http/2… 19:45:02 (it got quiet) 19:46:49 So we're aiming for the enterprise approach 19:47:14 = if you want security, add it somewhere else ;-) 19:48:17 (For reference, I totally agree - munin-httpd is not the place for these mechanisms) 19:48:18 i might add some filtering, to choose the nodes you want to see 19:48:57 for security, listen on a UNIX domain socket (in the future) and proxy to that 19:49:04 which might just be vhost based, as it's tje easiest 19:49:12 Some views could be nice, and then those views could be limited in the proxying httpd. 19:49:18 that's planed, yes. 19:49:31 Cue: rejoicing 19:49:32 not a bad idea, really. Easy to configure views based on user or vhost. 19:50:05 if user then http basic 19:50:17 just pick a really hard-to-use configuration format to keep our reputation intact. m4 anyone? 19:50:42 json-based m4. 19:51:32 xslt for enterprise readiness 19:51:33 just pick a http header to trust from the forwarding proxy. Then you instantly get x509 and kerberos authentication and authorization for graph access. That is enterpricy :D 19:51:43 xslt of course, how could I miss that 19:52:00 brain deleted it in self-defense 19:52:47 k, solved 19:53:16 :) 19:56:43 note that i saw a product that configured itself with xslt. defaults where provided as the input xml, and an identity xslt. 19:57:33 (obviously java based) 19:58:00 anyway... next step is 2.1.12 19:58:19 So 2.1.12 will be java based 19:58:21 and ssm & I wrote some issues on GH 19:58:22 *make notes* 19:58:27 (Sorry, I will stop now) 19:58:33 ;) 19:59:06 i think it's a great way to use gh. 19:59:34 all the gh interactions are now notfied here on irc. 19:59:42 Hey! Don't joke about Java, ok? :D 19:59:45 (even spammy ones) 20:00:21 chteuchteu: *if* rewrite then *C*. 20:00:52 bikeshed! :) 20:01:38 ... but many did already write partial rewritrs in various languages, on various licensing terms 20:02:07 Do they really work? 20:02:28 i don't think any got much hype, but it means 2 things : 20:02:42 - the ecosystem is real 20:03:08 - there have all some interesting features to steal ;) 20:04:00 chteuchteu: they are all mostly maas. (muni as a service) 20:05:23 ok! 20:05:55 ... note that the dynazoom was our killer feature, and did some harm to them, as it's quite difficult to replicate efficiently. 20:06:21 (i struggled enugh to agree) 20:06:49 multigraph was the second. now that 2.1 heavily uses it. 20:07:45 some feature where never implemebted in copycats. 20:08:15 such as .sum and .total 20:08:23 .cdef? 20:08:36 (which are missing from 2.1) 20:09:06 cdef has some alternates, .sum has none. 20:09:28 .cdef works in 2.1 20:10:02 i'd like to rewrtiz the only ruby plugin 20:10:09 (tomcat) 20:10:54 i'd like to rewrtie the py based also, to limit deps, but it's less an issue. 20:12:07 any c based plugins in use? 20:12:11 py is used at *build* time for the guide. but should not be required at *runtime* 20:12:58 my multi1sec rewrtie in C got some attention ( i recieved 5 emails abojt it ) 20:13:07 for plugins in contrib, requirements are not as strict :) 20:14:19 indeed 20:14:48 munin-run-c is missint 20:15:15 About multicpu1sec, how comes it isn't available by default? It should, it is great! 20:15:37 It's my favorite plugin. It has great colors for screenshots! 20:16:55 #action TheSnide will *promote* the perl version of multi1sec to core, along with if1sec_ 20:17:49 i'm wondering about a cpu1sec, diskstat1sec & memory1sec. 20:18:02 interpreter stats: https://gist.github.com/ssm/3e7b9cdfd66feb480b56 20:18:36 having system 1sec plugin in core would be a nice marketing move for 2.1 20:19:13 perl -w :( 20:19:25 yup 20:19:31 lots of low hanging fruit in there 20:19:39 moving them to "use warnings" ? 20:19:45 yes 20:20:15 and /usr/local/bin/bash o_O 20:20:23 that's the HP-UX plugins 20:20:29 ... in core ;) 20:20:29 bash lives there when installed 20:20:41 oh, makes sense ;) 20:21:00 * ssm has no hp-ux machine to test those on 20:21:30 * be0rn should get his SunBlade up and running 20:22:14 Then again, I guess a VM could take care of that. 20:37:48 so, i ran out of subjects 20:38:08 I actually bought myself a new Raspberry Pi 2 20:38:30 So I'm planning to work on pages layout soon 20:38:37 (making things more responsive mainly) 20:40:41 That's it for me :) 20:41:25 * ssm has nothing else to mention 20:42:02 nice ;) 20:42:15 ok 20:42:26 #endmeeting