Percona cluster doesn't establish, crm errors
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
percona-cluster (Juju Charms Collection) |
Fix Released
|
High
|
Liam Young |
Bug Description
On node 0 (notice 15:37:00):
2015-05-06 15:37:00 INFO unit.mysql/
2015-05-06 15:37:18 WARNING unit.mysql/
2015-05-06 15:37:18 DEBUG unit.mysql/
2015-05-06 15:37:18 DEBUG unit.mysql/
2015-05-06 15:37:19 WARNING unit.mysql/
On node 1 (Notice 15:36:01):
2015-05-06 15:34:25 INFO unit.mysql/
2015-05-06 15:34:25 INFO unit.mysql/
2015-05-06 15:34:25 INFO unit.mysql/
2015-05-06 15:35:30 WARNING unit.mysql/
2015-05-06 15:35:30 INFO unit.mysql/
2015-05-06 15:35:32 INFO unit.mysql/
2015-05-06 15:35:36 INFO unit.mysql/
2015-05-06 15:35:42 INFO unit.mysql/
2015-05-06 15:35:50 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 INFO unit.mysql/
2015-05-06 15:36:01 ERROR juju.worker.
Node 2 is similar (15:36:03 failure timestamp)
Related branches
- Adam Collard (community): Needs Fixing
- James Page: Approve
-
Diff: 112 lines (+63/-0)2 files modifiedcharmhelpers/contrib/hahelpers/cluster.py (+25/-0)
tests/contrib/hahelpers/test_cluster_utils.py (+38/-0)
- Billy Olsen: Approve
-
Diff: 326 lines (+122/-19)7 files modifiedhooks/charmhelpers/contrib/hahelpers/cluster.py (+25/-0)
hooks/charmhelpers/contrib/peerstorage/__init__.py (+2/-0)
hooks/charmhelpers/core/hookenv.py (+86/-10)
hooks/charmhelpers/core/host.py (+1/-1)
hooks/charmhelpers/core/services/base.py (+2/-2)
hooks/charmhelpers/fetch/__init__.py (+1/-1)
hooks/percona_hooks.py (+5/-5)
- Liam Young (community): Approve
-
Diff: 98 lines (+36/-6)2 files modifiedcharmhelpers/contrib/hahelpers/cluster.py (+12/-3)
tests/contrib/hahelpers/test_cluster_utils.py (+24/-3)
- James Page: Approve
-
Diff: 96 lines (+29/-4)2 files modifiedhooks/charmhelpers/contrib/hahelpers/cluster.py (+25/-0)
hooks/percona_hooks.py (+4/-4)
tags: | added: openstack |
summary: |
- not waiting long enough for cluster lead to establish? + Percona cluster doesn't establish, crm errors |
Changed in percona-cluster (Juju Charms Collection): | |
status: | New → Confirmed |
Changed in percona-cluster (Juju Charms Collection): | |
status: | Confirmed → In Progress |
Changed in percona-cluster (Juju Charms Collection): | |
assignee: | nobody → Liam Young (gnuoy) |
milestone: | none → 15.07 |
Changed in percona-cluster (Juju Charms Collection): | |
status: | Fix Committed → Fix Released |
tags: | added: kanban-cross-team |
Dave, this is often a result of the corosync resource used to determine leader dithering across units during deploy time (which I assume is the case for you). A couple of points here; firstly, this problem will simple go away once we have juju leadership election supported in all the charms (since we will no longer rely on a corosync resource). Second, can ypu try setting the hacluster cluster_count option to the number of units in your mysql cluster. This will prevent corosync from reaching quorum until the specified number of units have joined the cluster. In practice that has drastically reduce the incidence of this kind of issue for me. Hope that helps.