Back to bug 1311362

Who When What Removed Added
Red Hat Bugzilla Rules Engine 2016-02-24 04:07:45 UTC Keywords ZStream
Byreddy 2016-02-24 06:39:40 UTC Doc Type Bug Fix Known Issue
Red Hat Bugzilla 2016-02-24 06:39:40 UTC Doc Type Known Issue Bug Fix
Bhaskarakiran 2016-02-24 07:27:41 UTC CC byarlaga
Blocks 1268895
Anuradha 2016-02-24 08:34:45 UTC CC atalur
Assignee pkarampu atalur
Doc Text Cause:
A new directory was added for glusterfs internal maintenance. When 'gluster volume heal <volname> info is run from nodes with new glusterfs, they are expecting this directory to be present but it isn't on older nodes. This leads projection of a misleading information on the output of heal info 'Failed to process entries completely', even though all the entries were processed.

Consequence:
'gluster v heal <volname> info' gives wrong information that it failed.

Workaround (if any):
Gather the heal info output from nodes with old-version of gluster. Or, check that no index entry other than xattrop-* is there in <brickpath>/.glusterfs/indices/xattrop/ in every brick.

Result:
Correct heal information is obtained.
Doc Type Bug Fix Known Issue
Anuradha 2016-02-25 09:07:13 UTC Doc Text Cause:
A new directory was added for glusterfs internal maintenance. When 'gluster volume heal <volname> info is run from nodes with new glusterfs, they are expecting this directory to be present but it isn't on older nodes. This leads projection of a misleading information on the output of heal info 'Failed to process entries completely', even though all the entries were processed.

Consequence:
'gluster v heal <volname> info' gives wrong information that it failed.

Workaround (if any):
Gather the heal info output from nodes with old-version of gluster. Or, check that no index entry other than xattrop-* is there in <brickpath>/.glusterfs/indices/xattrop/ in every brick.

Result:
Correct heal information is obtained.
Cause:
A new directory, '<path-to-brick>/.glusterfs/indices/dirty/', was added for glusterfs internal maintenance. When 'gluster volume heal <volname> info is run from nodes with new glusterfs, they are expecting this directory to be present but it isn't on older nodes. This leads projection of a misleading information on the output of heal info 'Failed to process entries completely', even though all the entries were processed.

Consequence:
'gluster v heal <volname> info' gives wrong information that it failed.

Workaround (if any):
Gather the heal info output from nodes with old-version of gluster. Or, check that no index entry other than xattrop-* is there in <brickpath>/.glusterfs/indices/xattrop/ in every brick.

Result:
Correct heal information is obtained.
Laura Bailey 2016-02-26 01:19:54 UTC Doc Text Cause:
A new directory, '<path-to-brick>/.glusterfs/indices/dirty/', was added for glusterfs internal maintenance. When 'gluster volume heal <volname> info is run from nodes with new glusterfs, they are expecting this directory to be present but it isn't on older nodes. This leads projection of a misleading information on the output of heal info 'Failed to process entries completely', even though all the entries were processed.

Consequence:
'gluster v heal <volname> info' gives wrong information that it failed.

Workaround (if any):
Gather the heal info output from nodes with old-version of gluster. Or, check that no index entry other than xattrop-* is there in <brickpath>/.glusterfs/indices/xattrop/ in every brick.

Result:
Correct heal information is obtained.
Cause:
A new directory, '<brickpath>/.glusterfs/indices/dirty/', was added for glusterfs internal maintenance. When 'gluster volume heal <volname> info is run from nodes with new glusterfs, they are expecting this directory to be present but it isn't on older nodes. This leads projection of a misleading information on the output of heal info 'Failed to process entries completely', even though all the entries were processed.

Consequence:
'gluster v heal <volname> info' gives wrong information that it failed.

Workaround (if any):
Gather the heal info output from nodes with old-version of gluster. Or, check that no index entry other than xattrop-* is there in <brickpath>/.glusterfs/indices/xattrop/ in every brick.

Result:
Correct heal information is obtained.
Rahul Hinduja 2016-05-02 09:19:22 UTC CC rhinduja
Blocks 1311817
QA Contact storage-qa-internal bsrirama
Rahul Hinduja 2016-05-02 09:24:02 UTC Blocks 1311817 1299184
Alok 2016-05-02 09:32:47 UTC CC asrivast
Red Hat Bugzilla Rules Engine 2016-05-02 10:20:24 UTC Target Release --- RHGS 3.1.3
Pranith Kumar K 2016-05-03 05:38:38 UTC Status NEW ASSIGNED
CC pkarampu
Anuradha 2016-05-04 06:23:59 UTC Blocks 1332798
Pranith Kumar K 2016-05-04 06:42:12 UTC Status ASSIGNED POST
Anuradha 2016-05-05 05:18:19 UTC Blocks 1333239
Anuradha 2016-05-05 05:27:46 UTC Blocks 1333243
Pranith Kumar K 2016-05-05 05:28:58 UTC Status POST MODIFIED
errata-xmlrpc 2016-05-10 11:26:30 UTC Status MODIFIED ON_QA
Milind Changire 2016-05-10 12:03:38 UTC Fixed In Version glusterfs-3.7.9-4
Byreddy 2016-05-12 09:02:13 UTC Status ON_QA VERIFIED
Anuradha 2016-05-13 11:00:18 UTC Doc Text Cause:
A new directory, '<brickpath>/.glusterfs/indices/dirty/', was added for glusterfs internal maintenance. When 'gluster volume heal <volname> info is run from nodes with new glusterfs, they are expecting this directory to be present but it isn't on older nodes. This leads projection of a misleading information on the output of heal info 'Failed to process entries completely', even though all the entries were processed.

Consequence:
'gluster v heal <volname> info' gives wrong information that it failed.

Workaround (if any):
Gather the heal info output from nodes with old-version of gluster. Or, check that no index entry other than xattrop-* is there in <brickpath>/.glusterfs/indices/xattrop/ in every brick.

Result:
Correct heal information is obtained.
Cause:
During in-service upgrade, old nodes won't have a particular glusterfs maintenance directory "<brickpath>/.indices/dirty/" which the heal info command expects to exist when run the new nodes.

Consequence:
As the directory is not found, even though the required information for healing is printed, heal info incorrectly reports failure.

Fix:
Ignore the missing directory in older nodes for heal info command.

Result:
Heal info command doesn't incorrectly fail.
Doc Type Known Issue Bug Fix
Laura Bailey 2016-06-03 04:15:07 UTC Doc Text Cause:
During in-service upgrade, old nodes won't have a particular glusterfs maintenance directory "<brickpath>/.indices/dirty/" which the heal info command expects to exist when run the new nodes.

Consequence:
As the directory is not found, even though the required information for healing is printed, heal info incorrectly reports failure.

Fix:
Ignore the missing directory in older nodes for heal info command.

Result:
Heal info command doesn't incorrectly fail.
The heal info command expects bricks to have the .indices/dirty directory. However, older nodes do not have this directory. In-service upgrades on these older nodes failed, despite all information required for in-service upgrade being present. The missing directory is now ignored on older nodes and in-service upgrades do not fail in this situation.
Flags needinfo?(atalur)
Anuradha 2016-06-03 07:01:22 UTC Doc Text The heal info command expects bricks to have the .indices/dirty directory. However, older nodes do not have this directory. In-service upgrades on these older nodes failed, despite all information required for in-service upgrade being present. The missing directory is now ignored on older nodes and in-service upgrades do not fail in this situation. The heal info command expects bricks to have the .indices/dirty directory. However, older nodes do not have this directory. After in-service upgrades on a node, heal info command failed when run from newer nodes due to missing .indices/dirty directory on old nodes. The missing directory is now ignored on older nodes and heal info command after in-service upgrades does not fail in this situation.
Flags needinfo?(atalur)
Laura Bailey 2016-06-06 00:20:51 UTC Doc Text The heal info command expects bricks to have the .indices/dirty directory. However, older nodes do not have this directory. After in-service upgrades on a node, heal info command failed when run from newer nodes due to missing .indices/dirty directory on old nodes. The missing directory is now ignored on older nodes and heal info command after in-service upgrades does not fail in this situation. The heal info command expects bricks to have the .indices/dirty directory. However, older nodes do not have this directory. After in-service upgrades on a node, the heal info command failed when run from newer nodes due to missing .indices/dirty directory on old nodes. The missing directory is now ignored on older nodes and heal info command after in-service upgrades does not fail in this situation.
errata-xmlrpc 2016-06-23 00:48:41 UTC Status VERIFIED RELEASE_PENDING
errata-xmlrpc 2016-06-23 05:09:06 UTC Status RELEASE_PENDING CLOSED
Resolution --- ERRATA
Last Closed 2016-06-23 01:09:06 UTC
Nag Pavan Chilakam 2016-07-18 14:58:18 UTC CC nchilaka
Rejy M Cyriac 2016-09-17 12:18:46 UTC CC rhs-bugs, storage-qa-internal
Component gluster-afr replicate
John Skeoch 2016-09-20 02:01:07 UTC CC smohan

Back to bug 1311362