Bug 1049766

Summary: DHT : Directory self heal is fixing layout for few Directories when Sub-volume comes online
Product: [Red Hat Storage] Red Hat Gluster Storage Reporter: Rachana Patel <racpatel>
Component: distributeAssignee: Nithya Balachandran <nbalacha>
Status: CLOSED DEFERRED QA Contact: storage-qa-internal <storage-qa-internal>
Severity: high Docs Contact:
Priority: medium    
Version: 2.1CC: grajaiya, mzywusko, spalai, vagarwal, vbellur
Target Milestone: ---   
Target Release: RHGS 2.1.2   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
: 1286143 1286144 (view as bug list) Environment:
Last Closed: 2015-11-27 11:54:04 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: 1286143, 1286144    
Attachments:
Description Flags
log
none
log none

Description Rachana Patel 2014-01-08 07:29:03 UTC 
Description of problem:
On DHT volume if User creates Directory when one or more sub-volume is down then self heal should create Directory on down sub volume when it comes online.
But it should not fix hash layout for the same Directory.
Only rebalance should fix layout.

Right now when Sub-volumes come back Directories are healed on down sub-volume and hash layout is being fixed for few directories(not on all and how it choose those Directory is not consistent)

Version-Release number of selected component (if applicable):
3.4.0.53rhs-1.el6rhs.x86_64

How reproducible:
always

Steps to Reproduce:
1. create DHT volume and mount using FUSE
2. bring down one or more subvolumes by killing brick process
3. create Directories from mount point
4. bring subvolume up by start force option.
5. from mount point execute commands which lookup for those directories
6. verify  hash layout for those directories in backend

Actual results:
Directory layout on sub-volume which was down

# file: rhs/brick1/b3/new20
trusted.gfid=0x3bcc2d8e106c4b37804ea4dee34566e6
trusted.glusterfs.dht=0x000000010000000099999999cccccccb        <-----------

# file: rhs/brick1/b3/new21
trusted.gfid=0x2bfeeeccc18b4e5bac002bc7e04c23d4
trusted.glusterfs.dht=0x00000001000000000000000000000000

# file: rhs/brick1/b3/new22
trusted.gfid=0xc10feccfcae448dd856d44436b9fd8bc
trusted.glusterfs.dht=0x00000001000000000000000000000000

# file: rhs/brick1/b3/new23
trusted.gfid=0x0afb356f1ea448deb32a211fd6049aad
trusted.glusterfs.dht=0x000000010000000099999999cccccccb      <-----------

# file: rhs/brick1/b3/new25
trusted.gfid=0x8be3eac8fc4b4c2da6b1c2158a7d37cf
trusted.glusterfs.dht=0x00000001000000000000000000000000


It is fixing hash layout for few Directories


Expected results:
Self heal should not fix hash layout of Directories

Additional info:
Comment 2 Rachana Patel 2014-01-08 07:51:44 UTC 
Created attachment 846997 [details]
log
Comment 4 Rachana Patel 2014-01-08 07:53:18 UTC 
Created attachment 846998 [details]
log
Comment 9 Shyamsundar 2014-01-16 07:56:32 UTC 
Testing for regression versus existing behaviour as in comment #6 and comment #7 in bug #1049771 was attempted and found the same behavior in BegBend and Corbett.

Looks like the root cause is the same for both bugs, but as scenarios are different we can continue to track them separately.
Comment 10 Gowrishankar Rajaiyan 2014-01-16 18:03:48 UTC 
Per 1/16 triage.
Comment 11 Vivek Agarwal 2014-01-17 07:37:55 UTC 
Per triage 1/16, Removing from the Corbett list
Comment 15 Susant Kumar Palai 2015-11-27 11:54:04 UTC 
Cloning this to 3.1. To be fixed in future.