Commit Graph

90 Commits

Author SHA1 Message Date
Joshua Boniface 63f4f9aed7 Lint: E722 do not use bare 'except' 2020-11-06 18:55:10 -05:00
Joshua Boniface 56ba7b1457 Bump version to 0.9.1 2020-10-29 12:16:38 -04:00
Joshua Boniface ec0b8acf90 Support per-VM migration type selectors
Allow a VM to specify its migration type as a default choice. The valid
options are "default" (i.e. behave as now), "live" which forces a live
migration only, and "shutdown" which forces a shutdown migration only.
The new option is treated as a VM meta option and is set to default if
not found.
2020-10-29 12:01:29 -04:00
Joshua Boniface 5d08ad9573 Fix incorrect keepalive interval setting 2020-10-26 11:44:45 -04:00
Joshua Boniface 0f299777f1 Modify version to 3-digit numbering
I expect 0.9 will be fairly long-lived, so add another decimal place so
I may continue adding tweaks to it.

THIS IS NOT SEMVER.
2020-10-26 02:13:11 -04:00
Joshua Boniface 890023cbfc Make sender wait dynamic based on receiver 2020-10-21 14:43:54 -04:00
Joshua Boniface 28abb018e3 Improve some timeouts and conditionals 2020-10-21 12:00:10 -04:00
Joshua Boniface 017953c2e6 Move lock release to phase D 2020-10-21 11:07:01 -04:00
Joshua Boniface 82b4d3ed1b Add missing prefix statements to loggers 2020-10-21 10:52:53 -04:00
Joshua Boniface bae366a316 Add waits and only receive check on send 2020-10-21 10:43:42 -04:00
Joshua Boniface 351076c15e Check if node changed during final check
Avoids situations where two migrates, to different nodes, happen in
rapid succession. Aborts the migration if the current target node no
longer matches what was set at the start of the execution.
2020-10-21 02:52:36 -04:00
Joshua Boniface 42514b9a50 Improve messages further 2020-10-21 02:41:42 -04:00
Joshua Boniface 611e47f338 Add messages to migration aborts
Results in some information duplication, but ensures logging of the
reason a migration was aborted separate from the error(s) this may
generate.
2020-10-21 02:38:42 -04:00
Joshua Boniface 1523959074 Move where setting last_ vars happens 2020-10-21 02:24:00 -04:00
Joshua Boniface ef762359f4 Adjust timing to avoid migrating to self quickly
Add another separate state lock, release it earlier, and ensure timings
are good to avoid double-migrating one VM.
2020-10-21 02:17:55 -04:00
Joshua Boniface 398d33778f Avoid stopping duplicates, just lock our own key 2020-10-20 16:10:39 -04:00
Joshua Boniface a6d492ed9f Remove spurious writes and adjust sleep 2020-10-20 16:04:26 -04:00
Joshua Boniface 11fa3b0df3 Remove additional wait and add last_node entries
These allow for aborting a migration to retain the previous settings and
override what the client set.
2020-10-20 15:58:55 -04:00
Joshua Boniface 442aa4e420 Tweak timers further 2020-10-20 15:43:59 -04:00
Joshua Boniface 3910843660 Add missing break 2020-10-20 15:39:29 -04:00
Joshua Boniface 70f3fdbfb9 Tweak the delays slightly on receive 2020-10-20 15:38:07 -04:00
Joshua Boniface 7cb0241a12 Attempt live migrates 3 times before proceeding 2020-10-20 15:33:41 -04:00
Joshua Boniface 9fb33ed7a7 Increase peer lock acquiring timers 2020-10-20 15:26:59 -04:00
Joshua Boniface abfe0108ab Better handle aborting migrations 2020-10-20 15:22:16 -04:00
Joshua Boniface 567fe8f36b Wait for existing migrations before proceeding 2020-10-20 15:12:32 -04:00
Joshua Boniface ec7b78b9b8 Add additional short sleep in receive 2020-10-20 13:29:17 -04:00
Joshua Boniface 224c8082ef Alter text of synchronization messages 2020-10-20 13:08:18 -04:00
Joshua Boniface f9e7e9884f Improve handling of VM migrations
The VM migration code was very old, very spaghettified, and prone to
strange failures.

Improve this by taking cues from the node primary migration. Use
synchronization between the nodes to ensure lockstep completion of the
migration in discrete steps.

A proper queue can be built later to integrate with this code more
cleanly.

References #108
2020-10-20 13:01:55 -04:00
Joshua Boniface 726501f4d4 Add additional logging to flush selector
Adds additional debug logging to the flush selector to determine how any
why any given node is selected. Useful for troubleshooting strange
choices.
2020-10-20 12:34:18 -04:00
Joshua Boniface c6e34c7dc6 Bump base version to 0.9 2020-10-18 14:31:19 -04:00
Joshua Boniface f749633f7c Use provisioned memory for mem migration selector
Use the new "provisioned" memory field, instead of the "allocated"
memory field, to determine the optimal node when using the "mem"
migration selector. This will take into account non-running VMs in the
calculation as well as running VMs.
2020-10-18 14:17:15 -04:00
Joshua Boniface a4b80be5ed Add provisioned memory to node info
Adds a separate field to the node memory, "provisioned", which totals
the amount of memory provisioned to all VMs on the node, regardless of
state, and in contrast to "allocated" which only counts running VMs.

Allows for the detection of potential overprovisioned states when
factoring in non-running VMs.

Includes the supporting code to get this data, since the original
implementation of VM memory selection was dependent on the VM being
running and getting this from libvirt. Now, if the VM is not active, it
gets this from the domain XML instead.
2020-10-18 14:17:15 -04:00
Joshua Boniface aa5f8c93fd Entirely disable IPv6 on bridged interfaces
Prevents any potential leakage due to autoconfigured IPv6 on bridged
interfaces. These are exclusively VM-side bridges, and the PVC host
should not have any IPv6 configuration on them, ever.
2020-10-15 11:00:59 -04:00
Joshua Boniface 9366977fe6 Copy d_domain before iterating
Prevents a bug where the thread can crash due to a change in the
d_domain object while running the for loop. By copying and iterating
over the copy, this becomes safer.
2020-09-16 15:12:37 -04:00
Joshua Boniface 65b44f2955 Avoid breaking keepalive during incoming migration
The keepalive was getting stuck gathering memoryStats from the
non-running VM, since it was in a paused state. Avoid this by just
skipping past the rest of the stats gathering if the VM isn't running.
2020-08-28 01:47:36 -04:00
Joshua Boniface 78dec77987 Bump version to 0.8 2020-08-26 10:24:44 -04:00
Joshua Boniface 921e57ca78 Fix syntax error 2020-08-20 23:05:56 -04:00
Joshua Boniface 3cc7df63f2 Add configurable VM shutdown timeout
Closes #102
2020-08-20 21:26:12 -04:00
Joshua Boniface e8e65934e3 Use logger prefix for thread debug logs 2020-08-17 14:30:21 -04:00
Joshua Boniface 24fda8a73f Use new debug logger for DNS Aggregator 2020-08-17 14:26:43 -04:00
Joshua Boniface 9b3ef6d610 Add connect timeout to Ceph
This doesn't seem to actually do anything (like most of these
timeouts...) but add it just for posterity.
2020-08-17 13:58:14 -04:00
Joshua Boniface b451c0e8e3 Add additional start/finish debug messages 2020-08-17 13:11:03 -04:00
Joshua Boniface f9b126a106 Make zkhandler accept failures more robustly
Most of these would silently fail if there was e.g. an issue with the ZK
connection. Instead, encase things in try blocks and handle the
exceptions in a more graceful way, returning None or False if
applicable. Except for locks, which should retry 5 times before
aborting.
2020-08-17 13:03:36 -04:00
Joshua Boniface 553f96e7ef Use logger for debug output
Using simple print statements was annoying (lack of timing info and
formatting), so move to using the debug logger for these instead with a
custom state ('d') with white text to differentiate them. Also indicate
which subthread of the keepalive each task is being executed in for
easier tracing of issues.
2020-08-17 12:46:52 -04:00
Joshua Boniface 65add58c9a Properly properly handle issue 2020-08-16 11:38:39 -04:00
Joshua Boniface 0a01d84290 Tie fence timers to keepalive_interval
Also wait 2 full keepalive intervals after fencing before doing anything
else, to give the Ceph cluster a chance to recover.
2020-08-15 12:38:03 -04:00
Joshua Boniface 4afb288429 Properly handle missing domain_name fail 2020-08-15 12:07:23 -04:00
Joshua Boniface 985ad5edc0 Warn if fencing will fail
Verify our IPMI state on startup, and then warn if fencing will fail.
For now, this is sufficient, but in future (requires refactoring) we
might want to adjust how fencing occurs based on this information.
2020-08-13 14:42:18 -04:00
Joshua Boniface 0587bcbd67 Go back to manual command for OSD stats
Using the Ceph library was a disaster here; it had no timeout or way to
force it to continue, so keepalives would become stuck and trigger fence
storms. Go back to the manual osd dump command with a 2s timeout which
is far more reliable and can be adequately terminated if it runs long.
2020-08-12 22:31:25 -04:00
Joshua Boniface e0cb4a58c3 Ensure zk_listener is readded after reconnect 2020-08-11 12:46:15 -04:00