=== akhavr1 is now known as akhavr | ||
axw | wallyworld: back, ready when you are | 01:03 |
---|---|---|
axw | wallyworld: lost you | 02:03 |
anastasiamac | babbageclunk: ping? | 02:54 |
wallyworld | axw: so i'm at the neighbour's house - the dick next door cut my internet cable | 03:06 |
axw | wallyworld: :/ | 03:06 |
wallyworld | nfi how long they will take to repair it | 03:06 |
axw | wallyworld: I'm instrumenting all the things. adding a prometheus collector for the mgo stats atm, have added pprof profile for sockets and sessions. about to run it and see if anything pops out | 03:07 |
wallyworld | let's see | 03:07 |
anastasiamac | wallyworld: what did u do to ur neighbour to have cable cut? | 03:08 |
wallyworld | anastasiamac: hes a fool - he cut down a tree and let it fall across my cable - snap. lucky it missed the electrity connection | 03:09 |
axw | wallyworld: I have a controller with 1 machine, 1 unit. getting ~10 mgo ops sent per second. they're not mgo/txn ops tho | 03:12 |
wallyworld | interesting to see how many sasl logins | 03:17 |
wallyworld | i'm looking at that atm | 03:17 |
wallyworld | axw: data point https://bugs.launchpad.net/juju/+bug/1696739/comments/8 | 03:45 |
mup | Bug #1696739: mongodb reporting saslStart and isMaster as being slow <mongodb> <performance> <juju:Triaged> <https://launchpad.net/bugs/1696739> | 03:45 |
axw | wallyworld: hm ok, interesting | 03:46 |
wallyworld | doesn't really match with behaviour seen on site i don't think | 03:47 |
axw | wallyworld: I'm seeing saslCmd ops regularly, with just 1 machine and 1 unit | 03:47 |
wallyworld | hmmm | 03:48 |
axw | wallyworld: also a lot of iterator start/stops | 03:48 |
axw | wallyworld: that appears to be the way the txns.log watcher works | 03:49 |
wallyworld | axw: in the last 10 minutes, the number of sasl logins increased by approx 130, idle system | 03:49 |
axw | seems odd to start/stop watching a capped collection all the time? | 03:49 |
wallyworld | yeah | 03:49 |
wallyworld | same time socket login calls increased by 2000 | 03:49 |
wallyworld | the socket login calls are all pretty much lease and presence | 03:50 |
wallyworld | and so of the 2000, 130 drop through to require a sasl login | 03:50 |
wallyworld | axw: i left 2 more comments, i think we could target what presence does | 04:07 |
axw | wallyworld: did you try already? | 04:36 |
axw | making the presence change | 04:36 |
wallyworld | axw: not yet, gathering info. i'd like to understand why there are so many cache misses; i need to read john's notes properly | 04:38 |
anastasiamac | wpk: could u plz jump back into our standup? | 07:34 |
anastasiamac | babbageclunk: ^^ | 07:34 |
wpk | I'm here | 07:35 |
=== babbageclunk is now known as babbageclunk-afk | ||
cory_fu_ | It seems like this is from the initial AllWatcher response, which presumably includes the entire model, so I suppose it would depend on the model that you're connecting to. But it also seems like we should consider chunking those frames; that seems like it would be safer for the controller as well | 14:29 |
cory_fu_ | Oops, pasted that out of order | 14:29 |
cory_fu_ | Regarding https://github.com/juju/python-libjuju/issues/136 I'm wondering what the maximum expected / reasonable frame size from the Juju websocket would be? I don't feel entirely comfortable having no restriction whatsoever. | 14:29 |
cmars | morning juju, can i get reviews of a model migration bugfix? https://github.com/juju/juju/pull/7478 https://github.com/juju/juju/pull/7479 | 14:43 |
urulama_ | balloons: ^ this one is important for 2.2 :D | 14:44 |
balloons | urulama_, so 2.2-rc2 is already invalidated, nice | 14:53 |
urulama_ | balloons: well, this got discovered yesterday during migrations testing | 14:54 |
=== akhavr1 is now known as akhavr | ||
cmars | balloons, i'll merge 7478 and 7479 this afternoon after testing an actual 2.1 -> 2.2 migration | 18:30 |
balloons | cmars, ty | 18:33 |
cmars | whoa | 22:28 |
cmars | i just migrated a model | 22:28 |
cmars | what is this magic? | 22:28 |
cmars | balloons, fix for LP:#1696828 has landed in 2.2, will land in develop shortly | 22:46 |
mup | Bug #1696828: failure to migrate a model -- meter status "" is not valid <juju:In Progress by cmars> <juju 2.2:Fix Committed> <juju trunk:In Progress by cmars> <https://launchpad.net/bugs/1696828> | 22:46 |
=== mup_ is now known as mup |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!