[2.0a2] Detected deadlocks in postgres should be handled list a serialization failure
Bug #1558785 reported by
Blake Rouse
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
MAAS |
Fix Released
|
Critical
|
Blake Rouse |
Bug Description
When a deadlock is detected this normally occurs when another transaction is holding a lock longer than the default timeout set in postgres. MAAS regiond should do a better job of catching these exceptions and retrying the transaction. At the moment it will fail which can cause issues with that rack controller registering and other places through out MAAS.
We should also do work to make sure that areas where the lock is being held to long is reduced. That can be done setting the log_lock_waits in the postgres config. But that is another issue for another bug.
Related branches
lp:~blake-rouse/maas/fix-1558785
- Gavin Panella (community): Approve
-
Diff: 543 lines (+252/-30)6 files modifiedsrc/maasserver/middleware.py (+3/-3)
src/maasserver/tests/test_middleware.py (+11/-1)
src/maasserver/utils/orm.py (+78/-10)
src/maasserver/utils/tests/test_orm.py (+122/-9)
src/maasserver/utils/tests/test_views.py (+31/-0)
src/maasserver/utils/views.py (+7/-7)
Changed in maas: | |
status: | In Progress → Fix Committed |
Changed in maas: | |
status: | Fix Committed → Fix Released |
To post a comment you must log in.