Bug #4858
closedmon: doesn't necessarily call reset() during an election cycle
0%
Description
We need to call Monitor::reset() at some point during an election in order to guarantee consistency. However, we don't. It's mostly called via bootstrap(), but bootstrap() is only called if a lease or whatever times out, or if it notices a consistency issue later on.
This means that it's possible that only the monitor initiating an election calls reset(), which is generally bad.
This applies to both Bobtail and Cuttlefish, and is the underlying cause of the troubles with #4747.
Updated by Greg Farnum almost 11 years ago
- Subject changed from mon: doesn't call reset on peons in an election cycle to mon: doesn't necessarily call reset() during an election cycle
- Description updated (diff)
- Priority changed from Normal to Urgent
It's a bit more subtle than I'd initially described it.
Updated by Greg Farnum almost 11 years ago
- Status changed from New to Fix Under Review
wip-4858-reset[-bobtail]. Will run through a suite once it's up on gitbuilder.
Updated by Greg Farnum almost 11 years ago
- Status changed from Fix Under Review to 7
Sage says it's good!
Updated by Greg Farnum almost 11 years ago
- Status changed from 7 to Resolved
Ran tests against the rados and monitor regression suites in bobtail and in next.
Merged into next in 6ae9bbb5d03cb5695a4ebb7a3c20f729de1bd67a and bobtail in ea9c76b8f457d44e63551d47b6dd30bd16a56468.