Red Hat Bugzilla – Bug 849186
fix lease reacquire after failure
Last modified: 2014-01-12 19:53:57 EST
Description of problem:
The paxos code was doing some unnecessary host_id checks when the same
host re-acquired a lease after it was uncleanly shut down previously.
test: existing failure/recovery tests should continue to work the same
sanlock: fix paxos acquire host_id check
Version-Release number of selected component (if applicable):
Steps to Reproduce:
To test this code path, a host needs to be holding a lease, and then be reset/restarted in some way such that it does not record the lease as free on storage, i.e. the lease on disk still shows that the host holds it, but the host has in fact lost the lease. Once restarted, the host should be made to acquire the same lease. At this point it will see that it's still the owner on disk, although with a previous generation.
A couple ways you could cause a host to be restarted without releasing a lease on disk:
- Disconnect the storage holding the lease from the host.
- Reset power on the host while it's holding a lease.
storage disconnected from the host and reconnected after several minutes. the host acquired the same lease that it was holding before.
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.