Release notes for Gluster 3.12.2
This is a bugfix release. The release notes for 3.12.0, 3.12.1, 3.12.2 contain a listing of all the new features that were added and bugs fixed in the GlusterFS 3.12 stable release.
Major changes, features and limitations addressed in this release
1.) In a pure distribute volume there is no source to heal the replaced brick
from and hence would cause a loss of data that was present in the replaced brick.
The CLI has been enhanced to prevent a user from inadvertently using replace brick
in a pure distribute volume. It is advised to use add/remove brick to migrate from
an existing brick in a pure distribute volume.
Major issues
-
Expanding a gluster volume that is sharded may cause file corruption
- Sharded volumes are typically used for VM images, if such volumes are expanded or possibly contracted (i.e add/remove bricks and rebalance) there are reports of VM images getting corrupted.
- The last known cause for corruption #1465123 is still pending, and not yet part of this release.
-
Gluster volume restarts fail if the sub directory export feature is in use. Status of this issue can be tracked here, #1501315
-
Mounting a gluster snapshot will fail, when attempting a FUSE based mount of the snapshot. So for the current users, it is recommend to only access snapshot via ".snaps" directory on a mounted gluster volume. Status of this issue can be tracked here, #1501378
Bugs addressed
A total of 31 patches have been merged, addressing 28 bugs
- #1490493: Sub-directory mount details are incorrect in /proc/mounts
- #1491178: GlusterD returns a bad memory pointer in glusterd_get_args_from_dict()
- #1491292: Provide brick list as part of VOLUME_CREATE event.
- #1491690: rpc: TLSv1_2_method() is deprecated in OpenSSL-1.1
- #1492026: set the shard-block-size to 64MB in virt profile
- #1492061: CLIENT_CONNECT event not being raised
- #1492066: AFR_SUBVOL_UP and AFR_SUBVOLS_DOWN events not working
- #1493975: disallow replace brick operation on plain distribute volume
- #1494523: Spelling errors in 3.12.1
- #1495162: glusterd ends up with multiple uuids for the same node
- #1495397: Make event-history feature configurable and have it disabled by default
- #1495858: gluster volume create asks for confirmation for replica-2 volume even with force
- #1496238: [geo-rep]: Scheduler help needs correction for description of --no-color
- #1496317: [afr] split-brain observed on T files post hardlink and rename in x3 volume
- #1496326: [GNFS+EC] lock is being granted to 2 different client for the same data range at a time after performing lock acquire/release from the clients1
- #1497084: glusterfs process consume huge memory on both server and client node
- #1499123: Readdirp is considerably slower than readdir on acl clients
- #1499150: Improve performance with xattrop update.
- #1499158: client-io-threads option not working for replicated volumes
- #1499202: self-heal daemon stuck
- #1499392: [geo-rep]: Improve the output message to reflect the real failure with schedule_georep script
- #1500396: [geo-rep]: Observed "Operation not supported" error with traceback on slave log
- #1500472: Use a bitmap to store local node info instead of conf->local_nodeuuids[i].uuids
- #1500662: gluster volume heal info "healed" and "heal-failed" showing wrong information
- #1500835: [geo-rep]: Status shows ACTIVE for most workers in EC before it becomes the PASSIVE
- #1500841: [geo-rep]: Worker crashes with OSError: [Errno 61] No data available
- #1500845: [geo-rep] master worker crash with interrupted system call
- #1500853: [geo-rep]: Incorrect last sync "0" during hystory crawl after upgrade/stop-start
- #1501022: Make choose-local configurable through
volume-set
command - #1501154: Brick Multiplexing: Gluster volume start force complains with command "Error : Request timed out" when there are multiple volumes