19:27:36 <TheSnide> #startmeeting
19:27:36 <MeetBot> Meeting started Tue Jul 29 19:27:36 2014 UTC.  The chair is TheSnide. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:27:36 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
19:27:43 <TheSnide> hi all
19:32:51 <TheSnide> small weekly meeting
19:33:06 <TheSnide> #info munin-c
19:34:01 <TheSnide> #undo
19:34:01 <MeetBot> Removing item from minutes: <MeetBot.items.Info object at 0x13da8d0>
19:34:06 <TheSnide> #topic munin-c
19:34:34 <TheSnide> Some work is being done to make inetd.c work in win32
19:34:55 <TheSnide> (mostly for it to behave as a NT service)
19:35:22 <TheSnide> It will use the cygwin DLL for now, as it ease port.
19:35:55 <TheSnide> Then I'll create a WiX project in order to be able to distribute it as a MSI package.
19:36:36 <TheSnide> Also, there's a new (standalone) perfmon_ win32 plugin, that exports PerfMonitor counters to munin
19:38:00 <TheSnide> plan is to include it in the MSI. I don't know about integrating it directly in the munin-c repo, as it has very little value outside win32.
19:38:58 <TheSnide> A huge difficulty is that win32 is basically UNICODE only, whereas munin-c is ASCII only.
19:39:45 <TheSnide> #topic test coverage
19:40:39 <TheSnide> ssm made some nice commits relative to test coverage. I'm trying to use https://coveralls.io to show those automatically... with little succes for now.
19:41:13 <TheSnide> It is mostly because we have basically 3 projects in our repo: common, master & node.
19:41:39 <TheSnide> so, an unified coverage db cannot be used out-of-the-box.
19:41:59 <TheSnide> #topic 2.2
19:42:21 <TheSnide> I think i'll release 2.2 very soonish.
19:42:29 <be0rn> \o/
19:42:37 <Kvisle> \o/
19:43:07 <TheSnide> This would enable to have some sweetness out there, and begin to work on 2.3, which will be 2.4 (or 3.0)
19:43:40 <TheSnide> as it will feature the removal of all the datafile (text & storable).
19:43:49 <TheSnide> along with a persistent DB.
19:44:19 <TheSnide> Which removes all the single-threaded portions of the munin code.
19:44:34 <TheSnide> (that currently involve reading/writing the datafiles)
19:44:58 <TheSnide> It also enables a distributed master, along with a pg DB.
19:45:18 <TheSnide> .. but as it might take some time to land, i'd rather push 2.2 out.
19:46:08 <TheSnide> also, I'm releasing some small enhancements in 2.0, mostly in the async area.
19:47:45 <TheSnide> They are fixing some design failures. I don't like pushing enh in 2.0, but as 2.0.22 will be the version in jessie+1, I'd really like to have a rock-solid version.
19:48:02 <TheSnide> the enh are (so far) :
19:49:22 <TheSnide> - the use_node_name keyword can have a new value "ignore", which will be the default (I know... it's bad, but the current default : "use_node_name yes" is plain horrible)
19:50:39 <TheSnide> ... It will then just ask "list " to the node, ie asking without any node name. That will what ppl think.
19:51:04 <TheSnide> Since all the docs never mention "list $nodename" when debugging.
19:52:16 <TheSnide> - the async plugin plugin_rate will be stored in the SPOOL-META dictionary. It is text-based, whereas the current is DB_File based.
19:53:07 <TheSnide> also, i'll modify the defaults pruning values for async spools. Average size of 100MiB isn't nice IMHO.
19:53:40 <TheSnide> (remember, it's space used on *node*)
19:54:30 <TheSnide> #topic packaging
19:56:38 <TheSnide> I'd suggest packagers to use a dedicated munin-cgi package.
19:57:03 <TheSnide> That package can then be focused on _configuring_ munin for the various httpd servers.
19:57:31 <TheSnide> Also it should autostart fastcgi containers, and deprecated simple CGI.
19:58:45 <TheSnide> initd-style containers makes it _very_ easy to pre-configure and use
19:58:55 <TheSnide> #topic q&a
19:59:01 <TheSnide> any questions ?
19:59:29 <helmut> can you elaborate on the unicode difficulty?
19:59:52 <TheSnide> helmut: basically "isspace does not like unicode"
20:00:12 <helmut> is it that win32 tools emit non-ascii stuff?
20:00:59 <TheSnide> UTF-16 or UCS-2
20:01:17 <TheSnide> (it might be configurable)
20:01:32 * TheSnide didn't look closely at it yet.
20:02:00 <helmut> I can see why using UTF-16 would pose a problem, but I thought most of the world had moved to UTF-8 (where ASCII is a subset)
20:02:00 <Kvisle> windows provide ansi functions for most of its api -- I wouldn't recommend going that route, but it may be an option.
20:02:52 <TheSnide> also, win32 is mostly C++ and not C.
20:03:04 <TheSnide> (minor, but it adds)
20:03:08 <Kvisle> wa?
20:03:19 <helmut> thanks. nextup, I fail to see the functional distinction between the munin and munin-cgi packages
20:04:09 <TheSnide> Kvisle: i'm quite bad in w32, so it might be a coder-issue :)
20:04:29 <TheSnide> munin is munin-master
20:04:57 <TheSnide> munin-cgi is a package that switchs munin to cgi mode.
20:05:24 <TheSnide> (it is a little abuse of the package system, but I think it's OK)
20:05:31 <helmut> when do you want to use munin-master without munin-cgi?
20:10:43 <TheSnide> helmut: for very simple setups, it makes sense.
20:12:28 <TheSnide> http://msdn.microsoft.com/en-us/goglobal/bb688113.aspx
20:12:40 <TheSnide> (for Globalization Step-by-Step)
20:13:42 <TheSnide> if debian wants to make CGI default fine for me, but we got burn before.
20:14:59 <TheSnide> anything else ?
20:16:20 <TheSnide> #endmeeting