|
2015-06-16
§
|
| 05:56 |
<godog> |
bump ES replication throttling to 60mb/s |
[production] |
| 05:50 |
<manybubbles> |
ok - we're yellow and recovering. ops can take this from here. We have a root cause and we have things I can complain about to the elastic folks I plan to meet with today anyway. I'm going to finish waking up now. |
[production] |
| 05:49 |
<manybubbles> |
reenabling puppet agent on elasticsearch machines |
[production] |
| 05:46 |
<manybubbles> |
I expect them to be red for another few minutes during the initial master recovery |
[production] |
| 05:46 |
<manybubbles> |
started all elasticsearch nodes and now they are recovering. |
[production] |
| 05:41 |
<godog> |
restart gmond on elastic1007 |
[production] |
| 05:39 |
<filippo> |
Synchronized wmf-config/PoolCounterSettings-common.php: throttle ES (duration: 00m 13s) |
[production] |
| 05:25 |
<manybubbles> |
shutting down all the elasticsearch on the elasticsearch nodes against - another full cluster restart should fix it like it did last time............... |
[production] |
| 05:11 |
<godog> |
restart elasticsearch on elastic1031 |
[production] |
| 03:06 |
<springle> |
Synchronized wmf-config/db-eqiad.php: depool db1073 (duration: 00m 12s) |
[production] |
| 02:27 |
<LocalisationUpdate> |
completed (1.26wmf9) at 2015-06-16 02:27:51+00:00 |
[production] |
| 02:24 |
<l10nupdate> |
Synchronized php-1.26wmf9/cache/l10n: (no message) (duration: 05m 52s) |
[production] |
| 00:55 |
<tgr> |
running extensions/Gather/maintenance/updateCounts.php for gather wikis - https://phabricator.wikimedia.org/T101460 |
[production] |
| 00:52 |
<springle> |
Synchronized wmf-config/db-eqiad.php: repool db1057, warm up (duration: 00m 13s) |
[production] |
| 00:46 |
<godog> |
killed bacula-fd on graphite1001, shouldn't be running and consuming bandwidth (cc akosiaris) |
[production] |
| 00:27 |
<godog> |
kill python stats on cp1052, filling /tmp |
[production] |
|
2015-06-15
§
|
| 23:42 |
<ori> |
Cleaning up renamed jobqueue metrics on graphite{1,2}001 |
[production] |
| 23:01 |
<godog> |
killed bacula-fd on graphite2001, shouldn't be running and consuming bandwidth (cc akosiaris) |
[production] |
| 22:54 |
<hoo> |
Synchronized wmf-config/filebackend.php: Fix commons image inclusion after commons went https only (duration: 00m 14s) |
[production] |
| 22:18 |
<godog> |
run disk stress-test on restbase1007 / restbase1009 |
[production] |
| 22:06 |
<twentyafterfour> |
Synchronized hhvm-fatal-error.php: deploy: Guard header() call in error page (duration: 00m 15s) |
[production] |
| 22:05 |
<twentyafterfour> |
Synchronized wmf-config/InitialiseSettings-labs.php: deploy: Never use wgServer/wgCanonicalServer values from production in labs (duration: 00m 12s) |
[production] |
| 20:37 |
<yurik> |
Synchronized docroot/bits/WikipediaMobileFirefoxOS: Bumping FirefoxOS app to latest (duration: 00m 14s) |
[production] |
| 20:30 |
<godog> |
bounce cassandra on restbase1003 |
[production] |
| 20:18 |
<godog> |
start cassandra on restbase1008, bootstrapping |
[production] |
| 20:04 |
<godog> |
sign restbase1008 key, run puppet |
[production] |
| 20:00 |
<godog> |
powercycle restbase1007, investigate disk issue |
[production] |
| 19:07 |
<ori> |
Synchronized php-1.26wmf9/includes/jobqueue: 0a32aa3be4: jobqueue: use more sensible metric key names (duration: 00m 13s) |
[production] |
| 16:57 |
<thcipriani> |
Synchronized wmf-config/InitialiseSettings.php: SWAT: Grant cloudadmins the 'editallhiera' right [[gerrit:218115]] (duration: 00m 14s) |
[production] |
| 16:49 |
<thcipriani> |
Synchronized php-1.26wmf9/extensions/OpenStackManager/OpenStackManagerHooks.php: SWAT: refer to user the right way (duration: 00m 13s) |
[production] |
| 16:48 |
<godog> |
powercycle graphite1002, no ssh, unresponsive console |
[production] |
| 16:19 |
<jynus> |
upgrading es1005 mysql service while depooled |
[production] |
| 16:12 |
<thcipriani> |
Synchronized wmf-config/InitialiseSettings.php: SWAT: Grant cloudadmins the 'editallhiera' right [[gerrit:218115]] (duration: 00m 12s) |
[production] |
| 16:10 |
<bblack> |
pybal restarts complete, all ok |
[production] |
| 16:09 |
<thcipriani> |
Finished scap: SWAT: Openstack manager and language updates (duration: 21m 27s) |
[production] |
| 15:47 |
<thcipriani> |
Started scap: SWAT: Openstack manager and language updates |
[production] |
| 15:46 |
<bblack> |
starting pybal restart process for config changes ( https://gerrit.wikimedia.org/r/#/c/218285/ ), inactives first w/ manual verification of ok-ness |
[production] |
| 15:11 |
<bblack> |
rebooting cp3041 (downtimed) |
[production] |
| 15:00 |
<_joe_> |
ES is green |
[production] |
| 14:38 |
<aude> |
Synchronized php-1.26wmf9/extensions/Wikidata: Fix property label constraints bug (duration: 00m 24s) |
[production] |
| 14:27 |
<aude> |
Synchronized arbitraryaccess.dblist: Enable arbitrary access on s7 wikis (duration: 00m 13s) |
[production] |
| 13:47 |
<jynus> |
enabling puppet on all elastic* nodes, should enable also ganglia |
[production] |
| 13:11 |
<demon> |
Synchronized wmf-config/PoolCounterSettings-common.php: all the search (duration: 00m 12s) |
[production] |
| 13:04 |
<_joe_> |
re-scaling down the recovery index bandwidth in ES to 20 mb/s |
[production] |
| 12:52 |
<demon> |
Synchronized wmf-config/PoolCounterSettings-common.php: partially turn search back on (duration: 00m 13s) |
[production] |
| 11:54 |
<_joe_> |
raised the ES index replica bandwidth limit to 60mb |
[production] |
| 11:31 |
<akosiaris> |
migrating etherpad.wikimedia.org to etherpad1001.eqiad.wmnet |
[production] |
| 11:15 |
<_joe_> |
raised the max bytes for ES recovery to 40mbps |
[production] |
| 10:49 |
<manybubbles> |
and we're yellow right now. |
[production] |
| 10:49 |
<manybubbles> |
the initial primaries stage - the red stage of the rolling restart - recovers quick-ish |
[production] |