Bug 840737

Summary:	After the upgrade from glusterfs 3.2 to 3.3, the content of stipe and distributed-stripe is not available in mounted directory
Product:	[Community] GlusterFS	Reporter:	Rahul Hinduja <rhinduja>
Component:	stripe	Assignee:	shishir gowda <sgowda>
Status:	CLOSED WORKSFORME	QA Contact:
Severity:	medium	Docs Contact:
Priority:	medium
Version:	pre-release	CC:	amarts, gluster-bugs, nsathyan, vinaraya
Target Milestone:	---	Keywords:	Triaged
Target Release:	---
Hardware:	Unspecified
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:
Clones:	852581 (view as bug list)		Environment:
Last Closed:	2012-12-26 07:11:09 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:	852581

Description Rahul Hinduja 2012-07-17 06:43:35 UTC

Description of problem:

After performing the migration from 3.2 glusterfs to 3.3 glusterfs, tried mounting the stripe and distributed stiped volumes. The data is not available in the mount directory.

Version-Release number of selected component (if applicable):


How reproducible:


Steps to Reproduce:
1. Use glusterfs 3.2 and create the stripe and distributed-stripe volumes.
2. Mount these volumes and upload some data.
3. Upgrade the glusterfs version to 3.3
4. mount(using glusterfs or nfs) the stripe and distributed-stripe volume.
5. Data is not available to the mounted directory
  
Actual results:
Data is not available to the mounted directory

Expected results:

Data should be available

Additional info:
Confirmed that the data is available at the backend bricks.

Comment 1 Amar Tumballi 2012-12-24 09:39:02 UTC

Shishir, please mark this for a known issue (in migration guide) and then add the text.

----
if we make sure we pass on only 'readdir()' calls down and not 'readdirp()' calls, this should work out...

people should perform 'ls -l' and the distribute should have 'use-readdirp=no' option, with md-cache(stat-prefetch) disabled. need to test this, and see if it  works. If works, can close this with a doc update.
----

Comment 2 shishir gowda 2012-12-26 07:11:09 UTC

A work around to mention in the doc is as follows:

Before re-mounting the volume under 3.3
1. 'gluster volume set <volname> stat-prefetch off'
Mount with this xlator option enabled
2. --xlator-option *.use-readdirp=no
3. Run find . |xargs stat on the mount
4. After step 3 is completed, the volume can be used with stat-prefetch and doing a normal mount

We can't fix this issue in the current releases, as this is a migration from non-gfid backend to gfid based backend.