Bug 1227759
Summary: | Write performance from a Windows client on 3-way replicated volume decreases substantially when one brick in the replica set is brought down | ||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Product: | [Red Hat Storage] Red Hat Gluster Storage | Reporter: | Shruti Sampat <ssampat> | ||||||||||||
Component: | replicate | Assignee: | Ravishankar N <ravishankar> | ||||||||||||
Status: | CLOSED ERRATA | QA Contact: | Ben Turner <bturner> | ||||||||||||
Severity: | high | Docs Contact: | |||||||||||||
Priority: | high | ||||||||||||||
Version: | rhgs-3.1 | CC: | asriram, asrivast, bengland, bturner, divya, mlawrenc, nsathyan, ravishankar, rhs-bugs, rwheeler, storage-qa-internal, vagarwal | ||||||||||||
Target Milestone: | --- | Keywords: | ZStream | ||||||||||||
Target Release: | RHGS 3.1.1 | ||||||||||||||
Hardware: | Unspecified | ||||||||||||||
OS: | Unspecified | ||||||||||||||
Whiteboard: | |||||||||||||||
Fixed In Version: | glusterfs-3.7.1-13 | Doc Type: | Bug Fix | ||||||||||||
Doc Text: |
Previously, if a brick in a replica went down, there was a chance of drastic reduction in write speed due to extra fsyncs that happened. With the fix, this issue is resolved.
|
Story Points: | --- | ||||||||||||
Clone Of: | |||||||||||||||
: | 1250170 (view as bug list) | Environment: | |||||||||||||
Last Closed: | 2015-10-05 07:10:12 UTC | Type: | Bug | ||||||||||||
Regression: | --- | Mount Type: | --- | ||||||||||||
Documentation: | --- | CRM: | |||||||||||||
Verified Versions: | Category: | --- | |||||||||||||
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||||||||||
Cloudforms Team: | --- | Target Upstream Version: | |||||||||||||
Embargoed: | |||||||||||||||
Bug Depends On: | |||||||||||||||
Bug Blocks: | 1216951, 1223636, 1250170, 1251815, 1255698 | ||||||||||||||
Attachments: |
|
Description
Shruti Sampat
2015-06-03 12:42:09 UTC
Created attachment 1034293 [details]
profile info with all bricks up
Created attachment 1034295 [details]
profile info with one brick down - 1
Created attachment 1034297 [details]
profile info with one brick down - 2
Created attachment 1034298 [details]
profile info with one brick down - 3
Created attachment 1034300 [details]
profile info with one brick down - 4
I seem to remember that AFR started doing FSYNC on every write if a subvolume went down. In the past, most AFR volumes had only 2 subvolumes, so if you lost one of them, you were in danger of losing data if the 2nd one went down. However, with 3-way replication, it may be too pessimistic - you still have 2 out of 3 subvolumes so why start doing FSYNC at that point? suggested actions: run Gluster volume profile and see if FSYNC FOP is more frequent with 1 AFR subvolume down. See if FSYNC-per-WRITE code is still there in AFR and enabled. If so, can we can fix code to only do FSYNC-per-WRITE if we are down to a single AFR subvolume, regardless of whether we are doing 2- or 3-way replication. Is this really a blocker? I would suggest that this can slip out to an asynch update. Doc text is edited. Please sign off to be included in Known Issues. Hi Monti, I have made a slight modification to the doc text. Please update to this text. Verified on - glusterfs-3.7.1-15.el6rhs.x86_64 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. https://rhn.redhat.com/errata/RHSA-2015-1845.html |