Bug 1470566 - [RFE] Support changing from distribute to replicate with no active client operations
[RFE] Support changing from distribute to replicate with no active client ope...
Status: VERIFIED
Product: Red Hat Gluster Storage
Classification: Red Hat
Component: replicate (Show other bugs)
3.3
Unspecified Unspecified
medium Severity medium
: ---
: RHGS 3.4.0
Assigned To: Ravishankar N
Vijay Avuthu
: FutureFeature
Depends On:
Blocks: 1503132
  Show dependency treegraph
 
Reported: 2017-07-13 04:38 EDT by Sahina Bose
Modified: 2018-06-29 09:36 EDT (History)
7 users (show)

See Also:
Fixed In Version: glusterfs-3.12.2-1
Doc Type: Enhancement
Doc Text:
We are supporting converting from plain distribute to a distributed-replicate volume using 'gluster volume add-brick` command, provided there is no I/O happening from the client.
Story Points: ---
Clone Of:
Environment:
Last Closed:
Type: Bug
Regression: ---
Mount Type: ---
Documentation: ---
CRM:
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---


Attachments (Terms of Use)

  None (edit)
Description Sahina Bose 2017-07-13 04:38:28 EDT
Description of problem:

For one of the VM store use-cases, we want to start with a single node gluster volume and then later change to replica 3 gluster volume for redundancy.

This operation should be supported, and the heal should happen automatically when user sets a replica count and adds the additional bricks.


How reproducible:
NA
Comment 1 Atin Mukherjee 2017-07-18 00:27:23 EDT
T
Comment 3 Ravishankar N 2017-07-18 00:33:18 EDT
No, it should work fine, especially since https://bugzilla.redhat.com/show_bug.cgi?id=1276203 has been fixed.
Comment 4 Ravishankar N 2017-07-18 03:36:57 EDT
Realised that there is a corner case: BZ 1400092
Comment 7 Vijay Avuthu 2018-03-02 01:10:32 EST
Update:
=========

> tried below scenario

1) create  1 * 1 volume on a single node
2) create different IO which consists of
Comment 10 Vijay Avuthu 2018-03-12 06:28:49 EDT
Update :
============

> Covered below scenarios as well.

Scenario 1: converting 1 * 1 to 1 * 3

Steps:
 
1) create 1 * 1 distribute volume and start
2) create files and dirs from client ( 13k files and 13k dirs )
3) add bricks to volume so that it will convert to 1 * 3 ( Replicate )
4) Monitor the heal completion.
5) arequal-checksum is same on all the bricks and equals to arequal-checksum of client mountpoint


Scenario 1: converting 2 * 1 to 2 * 3

1) create 2 * 1 distribute volume and start
2) create files and dirs from client ( 13k files and 13k dirs )
3) add bricks to volume so that it will convert to 2 * 3 ( Distributed-Replicate )
4) Monitor the heal completion.
5) arequal-checksum is same on all the bricks in the replica set


Changing status to Verified.

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