mup | Bug #1647897 opened: Juju bootstrap proxy support <juju-core:New> <https://launchpad.net/bugs/1647897> | 01:40 |
---|---|---|
=== iatrou_ is now known as iatrou | ||
=== Tribaal_ is now known as Tribaal | ||
=== frankban|afk is now known as frankban | ||
gsamfira | hello folks. We have about 120 nodes constantly being deployed and redeployed by juju in a CI | 09:56 |
gsamfira | this environment has been online for a while now | 09:56 |
gsamfira | as a result, some collections have grown quite a bit | 09:56 |
gsamfira | one of them is: presence.beings | 09:56 |
gsamfira | juju:PRIMARY> db.presence.beings.count() | 09:56 |
gsamfira | 1646080 | 09:56 |
gsamfira | the other was logs.logs (which had about 14.000.000 entries) | 09:57 |
perrito666 | gsamfira: hey | 09:57 |
gsamfira | hey perrito666 | 09:57 |
perrito666 | gsamfira: logs should be rotated, something is wrong there | 09:57 |
gsamfira | I don;t think juju got a chance to rotate | 09:57 |
gsamfira | the db was under huge load | 09:58 |
perrito666 | gsamfira: but the rotation is done by juju | 09:58 |
perrito666 | and by rotation I mean just deletion | 09:58 |
gsamfira | mostly because of a bunch of queries against precence.being | 09:58 |
gsamfira | which had no index on model-uuid | 09:58 |
gsamfira | and did a COLLSCAN for every query | 09:58 |
perrito666 | gsamfira: this is 2.x? | 09:59 |
gsamfira | I can imagine. But juju kind of stopped working, erroring out with i/o error while talking to mongo | 09:59 |
gsamfira | 2.0.1 | 09:59 |
gsamfira | i/o error and i/o timeout | 09:59 |
gsamfira | I had to drop all connections to the state machine port, go into the database and create an index in presence.beings | 10:00 |
gsamfira | db.presence.beings.createIndex({"model-uuid": 1}) | 10:00 |
gsamfira | also on txns | 10:00 |
gsamfira | db.txns.createIndex({"s": 1}) | 10:00 |
gsamfira | just to get rid of a lot of spam in syslog about COLLSCANs | 10:00 |
perrito666 | thumper: babbageclunk any of you might be interested in this? | 10:01 |
gsamfira | also dropped all logs and saved 1.3 GB of disk space | 10:01 |
perrito666 | gsamfira: lol | 10:01 |
thumper | o/ | 10:01 |
perrito666 | so, for how long has this been running? | 10:01 |
gsamfira | the state machine is a 16 CPU core, 32 GB of RAM VM hosted on RAID10 10kRPM SAS disks | 10:01 |
thumper | gsamfira: logs is a capped collection so size bound | 10:01 |
thumper | presence is bollocks | 10:02 |
mup | Bug #1647897 changed: Juju bootstrap proxy support <juju-core:Invalid> <https://launchpad.net/bugs/1647897> | 10:02 |
thumper | and grows forever | 10:02 |
gsamfira | also might be worth mentioning we have enabled HA on this particular setup | 10:02 |
gsamfira | thumper: ouch | 10:02 |
gsamfira | 2 low hanging fruit we can implement is simply creating indexes on stuff we query | 10:02 |
gsamfira | especially if we query those frequently | 10:03 |
gsamfira | the environment has been up for a couple of months I think | 10:03 |
gsamfira | lemme check | 10:03 |
gsamfira | since october. Rougly 2 months | 10:05 |
gsamfira | perrito666: ^ | 10:05 |
perrito666 | gsamfira: tx that is a nice point of data | 10:06 |
gsamfira | http://paste.ubuntu.com/23592735/ <-- this might also be of interest | 10:06 |
gsamfira | this is a CI environment, a lot of units get torn down and spun up again | 10:06 |
gsamfira | so there is a lot of traffic | 10:06 |
gsamfira | perrito666: juju:PRIMARY> db.statuseshistory.count() | 10:08 |
gsamfira | 1810513 | 10:08 |
gsamfira | so you get an idea | 10:08 |
gsamfira | :) | 10:08 |
perrito666 | gsamfira: that is quite a reasonable size for status :) because its also capped but that one seems to be working | 10:09 |
gsamfira | yup. The environment seems stable again after creating the indexes and cleaning the logs | 10:10 |
perrito666 | gsamfira: that is good to know :) a quick workaround and low hanging fruit all together | 10:11 |
gsamfira | I'll create a patch for the indexes later this week | 10:12 |
mgz | voidspace: http://streams.canonical.com/juju/tools/agent/2.0.2/ | 10:17 |
mup | Bug #1357760 changed: ensure-availability (aka HA) should work with manual provider <cloud-installer> <ha> <landscape> <manual-provider> <manual-story> <juju:Triaged> <juju-core:Won't Fix> <https://launchpad.net/bugs/1357760> | 10:20 |
mup | Bug #1493058 changed: ensure-availability fails on GCE <docteam> <enable-ha> <gce-provider> <ha> <jujuqa> <juju:Invalid by thumper> <juju-core:Won't Fix> <juju-core 1.24:Won't Fix> <juju-core 1.25:Won't Fix> <https://launchpad.net/bugs/1493058> | 10:20 |
mup | Bug #1512569 changed: UniterSuite.TestRebootNowKillsHook fails with: uniter still alive <ci> <test-failure> <unit-tests> <juju:Triaged> <juju-core:Won't Fix> <https://launchpad.net/bugs/1512569> | 10:20 |
mgz | alexisb: can you backport your fix for bug 1631369 for 2.0 branch as well? | 10:20 |
mup | Bug #1631369: ExpireSuite.TestClaim_ExpiryInFuture_TimePasses took way too long <ci> <intermittent-failure> <regression> <unit-tests> <juju:In Progress by alexis-bruemmer> <https://launchpad.net/bugs/1631369> | 10:20 |
alexisb | mgz, sure | 10:21 |
axw_ | perrito666: I've created https://github.com/juju/juju/tree/feature-persistent-storage | 10:44 |
perrito666 | axw_: ack | 10:44 |
perrito666 | axw_: we should propose against that I presume | 10:44 |
axw_ | perrito666: yup, so we don't mess up 2.1 | 10:45 |
perrito666 | we dont mess up things, we awesome-ize them | 10:45 |
mup | Bug #1557726 changed: Restore fails on some openstacks like prodstack <backup-restore> <jujuqa> <openstack-provider> <juju:Fix Released by hduran-8> <juju 2.0:Fix Released by reedobrien> <juju-core:Won't Fix> <https://launchpad.net/bugs/1557726> | 10:47 |
=== akhavr1 is now known as akhavr | ||
natefinch | thumper: https://bugs.launchpad.net/juju/+bug/1648063 | 12:46 |
mup | Bug #1648063: kill-controller removes machines from migrated model <model-migration> <juju:Triaged> <https://launchpad.net/bugs/1648063> | 12:47 |
=== freyes__ is now known as freyes | ||
voidspace | jam: from within a hook context (debug-hooks) is there a way to get a list of all the bindings defined for the charm? | 12:56 |
voidspace | frobware: ^^^ do you know? | 12:57 |
jam | voidspace: generally it will be the things listed under "provides" or "requires" in charm-metadata.yaml | 13:00 |
frobware | voidspace: https://github.com/frobware/testcharms | 13:00 |
voidspace | jam: referring to the charm is what I was hoping to avoid | 13:01 |
voidspace | jam: but thanks | 13:01 |
voidspace | :-) | 13:01 |
rick_h | voidspace: :/ had hoped but nothing here of use https://jujucharms.com/docs/2.0/authors-hook-environment | 13:15 |
voidspace | rick_h: it would be a nice tool to have | 13:16 |
voidspace | rick_h: however... | 13:16 |
voidspace | rick_h: I have now torn down that environment and frobware has a test charm that I can deploy locally | 13:16 |
voidspace | rick_h: that defines useful stuff explicitly for playing with bindings | 13:16 |
rick_h | voidspace: yea, all good. Just noting that I can't find anything useful for what you were asking | 13:17 |
voidspace | rick_h: thanks for lookinh | 13:17 |
voidspace | rick_h: *looking | 13:17 |
voidspace | rick_h: but frobware has useful tools for playing with network-get | 13:17 |
natefinch | Review me? 2 line change: https://github.com/juju/juju/pull/6670 | 13:25 |
thumper | babbageclunk: https://github.com/juju/juju/pull/6669 | 13:56 |
=== jamespag` is now known as jamespage | ||
natefinch | perrito666: http://reports.vapour.ws/releases/4631/job/functional-backup-restore/attempt/4900 | 15:59 |
mup | Bug #1625624 changed: juju 2 doesn't remove openstack security groups <ci> <landscape> <openstack-provider> <sts> <juju:Fix Committed by gz> <juju 2.0:Fix Committed by gnuoy> <juju-core:Fix Released by gnuoy> <https://launchpad.net/bugs/1625624> | 16:50 |
=== frankban is now known as frankban|afk |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!