Status: New
Owner: ----
New issue 650 by [email protected]: gnt-instance activate-disks always
runs "drbdsetup syncer", even if syncer params are correct
http://code.google.com/p/ganeti/issues/detail?id=650
What steps will reproduce the problem?
1. Run gnt-instance activate-disks on a drbd instance whose disks are
already active
What is the expected output? What do you see instead?
Expected: no DRBD commands that change state should run
Observed: "drbdsetup syncer" command runs on the instance's drbd volume.
Please provide any additional information below.
Some context from an email conversation with thomasth:
"""
Assemble(...) calls SetSyncParams(...) regardless of the status of the
system. However, there is already a method in bdev.py
(DRBD8._SlowAssemble(...)) which tries to align the current state of the
DRBD device with the desired state, and calls DRBD8._AssembleNet(...) and
DRBD8._AssembleLocal(...) as appropriate.
We could simply move the call to SetSyncParams(...) from Assemble(...) to
_AssembleNet(...), so `drbdsetup syncer` would then only be called if the
network part of the DRBD device actually was re-established.
Changing DRBD syncer related disk parameters at the cluster or node group
level would then only be applied to the DRBD device once the instance is
rebooted, migrated or failed over (instead of by a mere gnt-instance
activate-disks, which can be done on the fly), but I think that's a
reasonable trade off.
"""
Not being familiar with the code, I can't tell for sure what the tradeoffs
would be. Of course it would be nice for an activate-disks to update syncer
parameters when the existing ones are not correct, but if that's not
possible I think it would still be best not to call drbdsetup when no setup
needs to be done.
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings
