Commit Graph

38 Commits

Author SHA1 Message Date
Joshua Boniface 46ffe352e3 Better handle subthread timeouts in keepalive
Prevent the main keepalive thread from getting stuck due to a subthread
taking an enormous time. If this happens, the rest of the main keepalive
will continue onward, thus ensuring that the main keepalive does not
fail for a significant number of cycles, which would cause a fence.
2020-08-11 11:37:26 -04:00
Joshua Boniface ccee124c8b Adjust fence failcount limit to 6 (30s)
The previous saving throw limit (3/15s) seems to have been too low. I
was observing bizarre failures where a node would be fenced while it was
still starting up. Some of this may have been related to Zookeeper
connections taking too long, but this was inconsistent.

Increase this to 6 saving throws (30s). This provides significantly more
time for a node to properly check in on startup before another node
fences it. In the real world, 15s vs 30s isn't that big of a downtime
change, but prevents false-positive fences.
2020-08-05 22:40:07 -04:00
Joshua Boniface 02343079c0 Improve fencing migrate layout
Open the option to do this in parallel with some threads
2020-08-05 22:26:01 -04:00
Joshua Boniface 37b83aad6a Add logging and use better conditional 2020-08-05 21:57:36 -04:00
Joshua Boniface 876f2424e0 Ensure dead state isn't written erroneously 2020-08-05 21:57:11 -04:00
Joshua Boniface 5871380e1b Avoid crashing VM stats thread if domain migrated 2020-06-10 17:10:46 -04:00
Joshua Boniface 654a3cb7fa Improve debug output and use ceph df util data 2020-06-06 22:52:49 -04:00
Joshua Boniface 9b65d3271a Improve handling of Ceph status gathering
Use the Rados library instead of random OS commands, which massively
improves the performance of these tasks.

Closes #97
2020-06-06 22:30:25 -04:00
Joshua Boniface 598b2025e8 Use Rados and add Ceph entries to pvcnoded.yaml 2020-06-06 21:12:51 -04:00
Joshua Boniface 70b787d1fd Move all VM functions into thread 2020-06-06 15:44:05 -04:00
Joshua Boniface e1310a05f2 Implement recording of VM stats during keepalive 2020-06-06 15:34:03 -04:00
Joshua Boniface 2ad6860dfe Move Ceph statistics gathering into thread 2020-06-06 13:25:02 -04:00
Joshua Boniface cebb4bbc1a Comment cleanup 2020-06-06 13:20:40 -04:00
Joshua Boniface a672e06dd2 Move fencing to end of keepalive function 2020-06-06 13:19:11 -04:00
Joshua Boniface 1db73bb892 Move libvirt closure into previous section 2020-06-06 13:18:37 -04:00
Joshua Boniface c1956072f0 Rename update_zookeeper function to node_keepalive 2020-06-06 12:49:50 -04:00
Joshua Boniface ce60836c34 Allow enforcement of live migration
Provides a CLI and API argument to force live migration, which triggers
a new VM state "migrate-live". The node daemon VMInstance during migrate
will read this flag from the state and, if enforced, will not trigger a
shutdown migration.

Closes #95
2020-06-06 12:00:44 -04:00
Joshua Boniface b5434ba744 Fix typo in variable name 2020-06-06 11:29:48 -04:00
Joshua Boniface b9e5b14f94 Update lastnode too if a self-migrate is aborted
References #92
2020-06-04 10:28:04 -04:00
Joshua Boniface 5d2031d99e Prevent a VM migrating to the same node
Prevents a rare edge case where a node can end up "migrating" to itself.
Quick hack to fix this, though like most of the VM management should
probably be rethought/rewritten later.

Fixes #92
2020-06-04 10:26:47 -04:00
Joshua Boniface 5f9836f96d Add error message to OSD parse fail 2020-05-12 11:04:38 -04:00
Joshua Boniface 95c59ba629 Improve flush handling slightly 2020-05-12 11:04:38 -04:00
Joshua Boniface 72a38fd437 Correct changed dhcp_reservations key name 2020-05-09 10:00:53 -04:00
Joshua Boniface b580760537 Add missing fmt_cyan variable 2020-05-08 18:15:02 -04:00
Joshua Boniface 331027d124 Add further tweaks to takeover state checks
Just ensure that everything is proper state before proceeding
2020-04-22 11:16:19 -04:00
Joshua Boniface 611e0edd80 Reorder last keepalive during cleanup
Make sure the stopping of the keepalive timer and final keepalive update
are done as the last step before complete shutdown. The previous setup
could conceivably result in a node being fenced should the cleanup
operations take longer than ~45 seconds, for instance if primary node
switchover took too long or blocked, or log watchers failed to stop
quickly enough. Ensures that keepalives will continue to be run during
the shutdown process until the last possible moment.
2020-04-12 03:49:29 -04:00
Joshua Boniface b413e042a6 Improve handling of primary contention
Previously, contention could occasionally cause a flap/dual primary
contention state due to the lack of checking within this function. This
could cause a state where a node transitions to primary than is almost
immediately shifted away, which could cause undefined behaviour in the
cluster.

The solution includes several elements:
    * Implement an exclusive lock operation in zkhandler
    * Switch the become_primary function to use this exclusive lock
    * Implement exclusive locking during the contention process
    * As a failsafe, check stat versions before setting the node as the
      primary node, in case another node already has
    * Delay the start of takeover/relinquish operations by slightly
      longer than the lock timeout
    * Make the current router_state conditions more explicit (positive
      conditionals rather than negative conditionals)

The new scenario ensures that during contention, only one secondary will
ever succeed at acquiring the lock. Ideally, the other would then grab
the lock and pass, but in testing this does not seem to be the case -
the lock always times out, so the failsafe check is technically not
needed but has been left as an added safety mechanism. With this setup,
the node that fails the contention will never block the switchover nor
will it try to force itself onto the cluster after another node has
successfully won contention.

Timeouts may need to be adjusted in the future, but the base timeout of
0.4 seconds (and transition delay of 0.5 seconds) seems to work reliably
during preliminary tests.
2020-04-12 03:40:17 -04:00
Joshua Boniface a671d9d457 Use consistent tense in messages 2020-04-08 22:00:51 -04:00
Joshua Boniface fee1c7dd6c Reorder cleanup and gracefully wait for flushes 2020-04-08 22:00:08 -04:00
Joshua Boniface 0a367898a0 Don't trigger aggregator fail if fine 2020-03-12 13:22:12 -04:00
Joshua Boniface c02bc0b46a Correct issues with VM lock freeing
Code was bad and using a depricated feature.
2020-03-02 12:45:12 -05:00
Joshua Boniface 1e4350ca6f Properly handle takeover state in VXNetworks
Most of these actions/conditionals were looking for primary state, but
were failing during node takeover. Update the conditionals to look for
both router states instead.

Also add a wait to lock flushing until a takeover is completed.
2020-03-02 10:41:00 -05:00
Joshua Boniface 57768f2583 Remove an obsolete script 2020-02-19 21:40:23 -05:00
Joshua Boniface e4e4e336b4 Handle invalid cursor setup cleanly
This seems to happen only during termination, so catch it and continue
so the loop terminates.
2020-02-19 16:29:59 -05:00
Joshua Boniface d2a5fe59c0 Use transitional takeover states for migration
Use a pair of transitional states, "takeover" and "relinquish", when
transitioning between primary and secondary coordinator states. This
provides a clsuter-wide record that the nodes are still working during
their synchronous transition states, and should allow clients to
determine when the node(s) have fully switched over. Also add an
additional 2 seconds of wait at the end of the transition jobs to ensure
everything has had a chance to start before proceeding.

References #72
2020-02-19 14:06:54 -05:00
Joshua Boniface 9c7041f12c Update package version to 0.7 2020-02-15 23:25:47 -05:00
Joshua Boniface 7ace5b5056 Remove /ceph/cmd pipe for (most) Ceph commands
Addresses #80
2020-02-08 23:40:02 -05:00
Joshua Boniface ce985234c3 Use consistent naming of components
Rename "pvcd" to "pvcnoded", and "pvc-api" to "pvcapid" so names for the
daemons are fully consistent. Update the names of the configuration
files as well to match this new formatting.

References #79
2020-02-08 19:34:07 -05:00