This repository has been archived by the owner on Mar 26, 2020. It is now read-only.
disable glustershd for GCS 1.0 #1485
Assignees
Labels
GCS/1.0
Issue is blocker for Gluster for Container Storage
Comments
|
Does this have an associated PR or plan for the next week in order to make 1.0? |
|
This PR #1490 disables the Self heal by default when a volume gets created. Currently a test is failing because it is expecting Self heal to be enabled by default, I will fix that and refresh that PR soon. |
ghost
removed
the
aravindavk
added a commit
that referenced
this issue
Jan 28, 2019
Xlator order for client volfile changed to, ``` debug/io-stats performance/read-ahead performance/io-threads performance/nl-cache performance/quick-read performance/open-behind performance/io-cache performance/readdir-ahead performance/write-behind performance/md-cache features/utime features/shard cluster/distribute ``` This PR also disables Self heal by default.(Fixes: #1485) Signed-off-by: Aravinda VK <[email protected]>
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Currently GD2 doesn't restart self heal daemon for every volume operation which is functionality wise incorrect. However we end up restarting self heal process on a container storage environment, with hundreds of RWX PVCs that's going to cause the performance penalty. PR #1460 attempted to do so and the results were not promising.
While the actual way to address this problem is to bring in multiplexing support to self heal daemon where with every volume operations the graph is attached/detached to self heal with out it being getting restarted but that work would require some more time before the GCS 1.0 timelines.
The mitigation plan is to disable self heal daemon and allow client self heal to ensure all the healing take place for the files which are accessed.
The text was updated successfully, but these errors were encountered: