19:28:30 <sumpfralle> #startmeeting
19:28:30 <MeetBot> Meeting started Wed Dec 19 19:28:30 2018 UTC.  The chair is sumpfralle. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:28:30 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
19:29:01 <sumpfralle> #chair TheSnide h01ger bcg chteuchteu kenyon be0rn ToxicFrog
19:29:01 <MeetBot> Current chairs: TheSnide ToxicFrog bcg be0rn chteuchteu h01ger kenyon sumpfralle
19:29:04 <sumpfralle> Welcome!
19:29:08 <sumpfralle> Who is around?
19:32:12 <ToxicFrog> I'm at work, so perhaps 10% availability.
19:33:54 <sumpfralle> ok - let us wait for others to join ...
19:34:07 <sumpfralle> maybe we should move even a bit more into the night :)
19:40:16 * h01ger is somewhat around
19:44:32 <TheSnide> hi
19:44:58 <sumpfralle> welcome!
19:45:06 <sumpfralle> so now we are a group :)
19:45:10 <sumpfralle> let us start ...
19:45:10 <TheSnide> sumpfralle: it was at 2130 CET before ;) ... and you asked to move it 1h earlier ;)
19:45:18 <sumpfralle> Did I?
19:45:28 <TheSnide> IIRC, yes
19:45:31 <sumpfralle> I think, it was someone else. I cannot tell.
19:45:39 <TheSnide> oh, well... maybe.
19:45:45 <sumpfralle> What is your favorite time, TheSnide?
19:45:52 <TheSnide> someone did. who exactly isn't really important
19:46:29 <TheSnide> 2100 would be a good compromise
19:46:44 <sumpfralle> This would be good for me, too.
19:46:45 <TheSnide> but, let's not change it.
19:46:50 <sumpfralle> ok :)
19:47:05 <sumpfralle> #topic What happened last week?
19:47:12 <TheSnide> let's say 2030, with a tolerance until 2100 ;)
19:47:19 <sumpfralle> good!
19:48:04 <TheSnide> last week, i was completely under water @ work... and will be until EOW.
19:48:21 <sumpfralle> poor you!
19:48:57 <TheSnide> will be in holidays starting of Saturday, so i really want to work on it this WE
19:49:09 <TheSnide> and, get the 3.0 out for the 25.
19:49:18 <sumpfralle> sounds great!
19:49:38 <TheSnide> it won't be a great release... but i don't really care anymore.
19:49:38 <sumpfralle> I did a bit of digging into the "use_node_name" issue - driven by ToxicFrog's deep dive into the code.
19:49:49 <sumpfralle> hehe - do not say this aloud :)
19:49:57 <TheSnide> bah
19:50:05 <sumpfralle> It will be a happy release and it will leave room for improvement ...
19:50:06 <TheSnide> it's _good enough_
19:50:13 <sumpfralle> fine
19:50:35 <TheSnide> 2 major things are missing:
19:50:37 <TheSnide> 1. limits
19:50:46 <sumpfralle> #topic Release for 3.0
19:50:59 <TheSnide> 2. why there's so many white stripes in dmm0
19:51:15 <sumpfralle> Yes, (2) is weird.
19:51:17 <TheSnide> limits should be very doable.
19:51:26 <TheSnide> 2. I have no clue whatsoever
19:51:43 <sumpfralle> I am tempted to finish the move to the VM donated to. I have the feeling it is a load issue.
19:51:59 <sumpfralle> (the host looks good - but the load numbers are unexpectedly high)
19:52:18 <sumpfralle> We will see afterwards (and before christmas).
19:53:13 <sumpfralle> We will do a "review the collection of plugins" session afterwards, or?
19:53:27 <sumpfralle> (after the release)
20:02:29 <hugin> [13munin] 15sumpfralle closed pull request #1138: Respect custom color for warning thresholds (06stable-2.0...06hrule-colour) 02https://git.io/fpFUm
20:02:51 <TheSnide> +1
20:02:57 <hugin> [13munin] 15sumpfralle closed issue #1135: warning-HRULES use wrong colour if individual colour for that field 02https://git.io/fp5wP
20:02:58 <hugin> [13munin] 15sumpfralle commented on issue #1135: It works for me - thus I merged the branch. 02https://git.io/fpj9w
20:03:29 <h01ger> hm, you didnt discuss a 2.0.44 release (date)
20:03:39 <sumpfralle> h01ger: in a minute :)
20:03:42 <h01ger> ah
20:03:48 <sumpfralle> TheSnide: anything else for the 3.0 release?
20:03:51 <TheSnide> the discussion, or the release ? ;)
20:04:14 <TheSnide> well, not really.
20:04:29 <sumpfralle> ok - so let us switch to stable ...
20:04:35 <sumpfralle> #topic Release 2.0.44
20:04:36 <TheSnide> I don't really like the deb packaging for 3.0
20:04:44 <TheSnide> ... But that's a whole other story
20:05:11 <sumpfralle> I am open for suggestions. We can do this after the release - it is still in experimental - thus nothing is set in stone.
20:05:33 <sumpfralle> Right now I am collecting last pieces for the 2.0.44 release.
20:05:51 <sumpfralle> I will tag it today.
20:08:49 <TheSnide> +1
20:09:14 <sumpfralle> Any other topics for today?
20:09:33 <TheSnide> ah, the docker version is working (if someone cares)
20:10:01 <h01ger> whats docker?
20:10:05 <h01ger> ;)
20:10:20 <TheSnide> h01ger: it's even based on debian-slim ;)
20:10:29 <sumpfralle> #topic Docker
20:10:35 <sumpfralle> sounds good!
20:10:45 <sumpfralle> Do you store the recipe somewhere?
20:10:52 <sumpfralle> Is it already hidden in the repository?
20:10:57 <TheSnide> i have to push it to the repo
20:11:03 <sumpfralle> good
20:11:13 <TheSnide> yeah, some is in the ./Dockerfile in munin.git
20:11:33 <TheSnide> I was thinking about using docker-compose for various parts
20:11:34 <sumpfralle> oh - I never looked at that one - good.
20:12:16 <TheSnide> and having only 1 process per container, as recommended. but i got lazy and i think that 1 container for the whole munin is nice.
20:12:32 <sumpfralle> sounds reasonable to me
20:12:32 <TheSnide> therefore there's cron + munin-httpd
20:12:38 <TheSnide> (inside)
20:12:53 <TheSnide> There's no munin-node, as it would be a little pointless
20:13:00 <h01ger> sounds reasonable to me too (to put more than 1 (related) process in the container)
20:13:11 <sumpfralle> Adding a timer for systemd would remove the need for a cron daemon, or?
20:13:18 <TheSnide> the _real_ issue being config
20:13:22 <sumpfralle> (just wild guessing)
20:13:27 <TheSnide> and state files
20:13:54 <TheSnide> "Adding a timer for systemd" Not really, as there's no systemd there ;)
20:14:03 <sumpfralle> also fine :)
20:14:09 <TheSnide> docker is basically "your process is PID 1"
20:14:29 <sumpfralle> ok
20:14:35 <TheSnide> so, munin-httpd is PID1 and cron is launched in bg
20:15:05 <TheSnide> well, it's obviously not PID1, but with the magic of PID namespaces it is.
20:15:59 <TheSnide> I was thinking about having munin-httpd running munin-update... but then I though that cron was quite fit for the task ;)
20:16:17 <sumpfralle> I like that.
20:16:34 <TheSnide> so, there's 2 _major_ pain points.
20:16:53 <TheSnide> which are related to the fact that containers are ephemeral. by design.
20:16:58 <sumpfralle> Configuration and data directories would be mounted from the outside, ode?
20:17:03 <sumpfralle> s/ode/or
20:20:48 <TheSnide> /etc/munin & /var/lib/munin are not persisted
20:21:15 <TheSnide> and, yes. They should be from the outside. Which is something that "leaks".
20:22:03 <TheSnide> also, the pgsql database isn't persisted.
20:22:31 <TheSnide> ... yeah, I use pg instead of sqlite, since well.. everything is prepackaged anyaway
20:29:25 <sumpfralle> Should we close the meeting for now and continue afterwards?
20:35:57 <sumpfralle> #endmeeting