Bug 1724184 - Thin-arbiter: SHD takes lock and inspects the state on TA during every index crawl
Summary: Thin-arbiter: SHD takes lock and inspects the state on TA during every index ...
Keywords:
Status: CLOSED NEXTRELEASE
Alias: None
Product: GlusterFS
Classification: Community
Component: replicate
Version: mainline
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Karthik U S
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks: 1729483
TreeView+ depends on / blocked
 
Reported: 2019-06-26 12:26 UTC by Karthik U S
Modified: 2019-07-12 11:49 UTC (History)
1 user (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
: 1729483 (view as bug list)
Environment:
Last Closed: 2019-07-12 11:19:15 UTC
Regression: ---
Mount Type: ---
Documentation: ---
CRM:
Verified Versions:


Attachments (Terms of Use)


Links
System ID Priority Status Summary Last Updated
Gluster.org Gerrit 22950 None Merged cluster/ta: Notify the clients only if there are pending heals 2019-07-12 11:19:14 UTC

Description Karthik U S 2019-06-26 12:26:53 UTC
Description of problem:

For thin arbiter volumes, even if there are no entries needing heal, SHD will take a lock on TA in AFR_TA_DOM_NOTIFY domain and gets the state and then unlocks. This will lead to invalidation of cached information on clients about the bricks state even though there is no pending heal. This happens during index crawl at every heal-timeout interval.

SHD should not take the lock on TA and invalidate the cached data on clients if there is no pending heal.

Comment 1 Worker Ant 2019-06-26 12:41:42 UTC
REVIEW: https://review.gluster.org/22950 (cluster/ta: Notify the clients only if there are pending heals) posted (#1) for review on master by Karthik U S

Comment 2 Worker Ant 2019-07-12 11:19:15 UTC
REVIEW: https://review.gluster.org/22950 (cluster/ta: Notify the clients only if there are pending heals) merged (#7) on master by Ravishankar N


Note You need to log in before you can comment on or make changes to this bug.