Back to bug 1043032

Who When What Removed Added
Matt Mahoney 2013-12-13 20:09:56 UTC Keywords ZStream
CC dtsang, knarra, mmahoney, pprakash, ssampat
Severity unspecified high
Dusmant 2013-12-16 09:10:14 UTC Status NEW ASSIGNED
Assignee dpati sabose
Flags needinfo?(mmahoney) needinfo?(mmahoney)
Sahina Bose 2013-12-16 09:39:51 UTC CC vbellur
Flags needinfo?(vbellur)
Sahina Bose 2013-12-16 09:41:54 UTC CC kaushal
Flags needinfo?(kaushal)
Kaushal 2013-12-16 10:38:18 UTC Flags needinfo?(mmahoney) needinfo?(mmahoney) needinfo?(vbellur) needinfo?(kaushal)
Sahina Bose 2013-12-17 11:53:10 UTC Status ASSIGNED POST
Link ID oVirt gerrit 22448
Sahina Bose 2013-12-18 02:13:49 UTC Priority unspecified high
Sahina Bose 2013-12-19 06:35:49 UTC Status POST MODIFIED
Sahina Bose 2013-12-20 06:45:04 UTC Status MODIFIED ON_QA
Fixed In Version cb13
Gowrishankar Rajaiyan 2013-12-23 04:31:18 UTC CC grajaiya
QA Contact sdharane mmahoney
Dusmant 2013-12-24 04:48:17 UTC Status ON_QA MODIFIED
CC dpati
Fixed In Version cb13
Assignee sabose kmayilsa
Kanagaraj 2013-12-31 06:13:06 UTC Status MODIFIED ON_QA
Fixed In Version CB14
Matt Mahoney 2014-01-03 14:14:31 UTC Status ON_QA VERIFIED
Kanagaraj 2014-01-06 07:11:42 UTC Doc Text Cause:
when adding a new host to an existing cluster, 'gluster peer probe' and 'gluster peer status' which is invoked by the sync job are executed in different servers one after the another immediately. RHSC always pick a random UP server whenever executing any gluster command. This results in related commands executed in different servers.

Consequence:
Newly added server gets removed from the Database by sync job as the new server was not found in the cluster. This is due to 'gluster peer status' was invoked on a different server.

Fix:
Same UP server will be used to execute all the gluster commands except distributed commands like 'volume status'. When adding a new host, the engine will wait for few seconds after 'gluster peer probe' is issued to get the same host appear in 'gluster peer status'.

Result:
Servers won't get removed by the sync job when they are added to an existing cluster.
Shalaka 2014-01-07 05:40:23 UTC CC kmayilsa, sharne
Doc Text Cause:
when adding a new host to an existing cluster, 'gluster peer probe' and 'gluster peer status' which is invoked by the sync job are executed in different servers one after the another immediately. RHSC always pick a random UP server whenever executing any gluster command. This results in related commands executed in different servers.

Consequence:
Newly added server gets removed from the Database by sync job as the new server was not found in the cluster. This is due to 'gluster peer status' was invoked on a different server.

Fix:
Same UP server will be used to execute all the gluster commands except distributed commands like 'volume status'. When adding a new host, the engine will wait for few seconds after 'gluster peer probe' is issued to get the same host appear in 'gluster peer status'.

Result:
Servers won't get removed by the sync job when they are added to an existing cluster.
Previously, newly added servers were getting removed from the Database by sync job as the new server was not found in the cluster. This was due to 'gluster peer status' being invoked on a different server. Now, with this update, while adding a new host, the engine waits for a few seconds after 'gluster peer probe' is issued to get the same host appear in 'gluster peer status' and the server in Up status will be used to execute all the gluster commands. The servers will not be removed from the sync job when they are added to an existing cluster.
Flags needinfo?(kmayilsa)
Shalaka 2014-01-09 06:54:23 UTC Doc Text Previously, newly added servers were getting removed from the Database by sync job as the new server was not found in the cluster. This was due to 'gluster peer status' being invoked on a different server. Now, with this update, while adding a new host, the engine waits for a few seconds after 'gluster peer probe' is issued to get the same host appear in 'gluster peer status' and the server in Up status will be used to execute all the gluster commands. The servers will not be removed from the sync job when they are added to an existing cluster. Previously, newly added servers were getting removed from the Database by sync job as the new server was not found in the cluster. This was due to 'gluster peer status' being invoked on a different server. Now, with this update, the servers will not be removed from the sync job when they are added to an existing cluster.
Jay Turner 2014-01-15 11:27:35 UTC Target Milestone --- RHS 2.1.2
Kanagaraj 2014-01-16 06:43:00 UTC Flags needinfo?(kmayilsa)
errata-xmlrpc 2014-02-25 05:42:06 UTC Status VERIFIED RELEASE_PENDING
errata-xmlrpc 2014-02-25 08:09:03 UTC Status RELEASE_PENDING CLOSED
Resolution --- ERRATA
Last Closed 2014-02-25 03:09:03 UTC
Rejy M Cyriac 2015-05-13 16:29:57 UTC Target Release --- RHGS 2.1.2
Target Milestone RHS 2.1.2 ---
John Skeoch 2016-04-18 10:06:41 UTC CC sankarshan

Back to bug 1043032