Merge lp:~raharper/curtin/trunk.md-retry-stop-resync into lp:~curtin-dev/curtin/trunk
Proposed by
Ryan Harper
Status: | Merged |
---|---|
Merged at revision: | 497 |
Proposed branch: | lp:~raharper/curtin/trunk.md-retry-stop-resync |
Merge into: | lp:~curtin-dev/curtin/trunk |
Diff against target: |
315 lines (+214/-23) 3 files modified
curtin/block/__init__.py (+0/-3) curtin/block/mdadm.py (+63/-10) tests/unittests/test_block_mdadm.py (+151/-10) |
To merge this branch: | bzr merge lp:~raharper/curtin/trunk.md-retry-stop-resync |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Server Team CI bot | continuous-integration | Approve | |
Scott Moser (community) | Approve | ||
Review via email:
|
Description of the change
mdadm_stop: Add retry and additional steps to halt a resync
In our vmtest (as well as seen in real deployments) after a reboot of a system with mdadm arrays already configured on disk, md may start a 'resync' operation to bring an array online. The resync operation sometimes prevents mdadm from stopping the array. This patch applies a few known methods to halt the resync, namely writing 'idle' to md/sync_action and changing the sync_max and sync_min values to 0.
Add additional unittests to exercise the retry path now taken to help ensure we can reliably shutdown a mdadm array even while in 'resync' state.
To post a comment you must log in.
Started a vmtest run here: /jenkins. ubuntu. com/server/ job/curtin- vmtest- devel-debug/ 43/
https:/