3.0 Update 4 Release Notes
Release Notes for Red Hat Storage - 3.0 Update 4
Chapter 1. Introduction
Red Hat Storage Server for On-Premise enables enterprises to treat physical storage as a virtualized, scalable, and centrally managed pool of storage by using commodity servers and storage hardware.
Red Hat Storage Server for Public Cloud packages GlusterFS as an Amazon Machine Image (AMI) for deploying scalable NAS in the AWS public cloud. This powerful storage server provides a highly available, scalable, virtualized, and centrally managed pool of storage for Amazon users.
Chapter 2. What's New in this Release?
- Support for Just a Bunch of Disks
- Red Hat Storage supports Just a Bunch Of Disks (JBOD) on servers with up to 24 disks. The complexity of RAID is avoided by using 3-way replication with JBOD.
- Three-way replication
- Synchronous three-way replication is now fully supported in Red Hat Storage volumes. Three-way replicated volumes do not require hardware or software RAID. Three-way replication yields best results when used in conjunction with JBODs. Quorum functionality on three-way replicated volumes helps to prevent split-brain scenarios. 3-way replicated volumes provide additional resilience in public cloud deployments as well.Three-way replication is also supported for Hadoop Deployments.
- Performance Enhancements
- Event threads: Two new performance enhancement options namely event-thread.client and event-thread.server are introduced. Configuring appropriate values minimizes the effect of slow throughput and response time in a Red Hat Storage trusted storage pool.
- Virtual Memory parameters: Two new virtual memory parameters can be configured to optimize performance of random, small file workloads and large file sequential workloads.
- Change in the location of the Gluster command log fileGluster commands executed on a node in the Red Hat Storage Truster Storage Pool is logged in the file
/var/log/glusterfs/cmd_history.log. It is no longer a hidden file.
- Samba-3 is deprecated from the Red Hat Storage 3.0 Update 4 releaseThe Samba version 3 is being deprecated from RHS 3.0 Update 4. Further updates will not be provided for Samba-3.x. It is recommended that you upgrade to Samba-4.x, which is provided in a separate channel or repository, for all updates including the security updates.RHN channel: rhel-x86_64-server-6-rh-gluster-3-sambaCDN repo: rh-gluster-3-samba-for-rhel-6-server-rpms
- The gluster CLI sub package is deprecated from the Red Hat Enterprise Linux optional channelThe
glusterfs-clisub-package is removed from the product listing for
rhel-optionalchannel or repository. No further versions of the sub-package will be shipped to the
rhel-optionalchannel or repository.
Chapter 3. Known Issues
3.1. Red Hat Storage
- BZ# 1202388After restoring a snapshot, attaching a new node to a Red Hat Storage Trusted Storage Pool fails due to checksum mismatch.This failure occurs in following scenarios:
Workaround: Ensure that you do not perform any file system operations on the snapshot that is about to be deleted.
- If quota is enabled on the volume and a snapshot is taken, the system does not recompute the quota checksum when the snapshot is restored. This leads to a checksum mismatch and the new node goes into a rejected state.Workaround:After snapshot restore,
glusterdservice must be restarted on existing nodes and the peer probe has to be executed on the new node.
- If there are different volume options configured in the origin volume and the snapshot volume, and a new node is added, subsequently the same snapshot is restored on the volume, the restored volume will have a different checksum in the new node. Restarting of any node in the Red Hat Storage Trusted Storage Pool results in the new node to move to peer-reject state.Workaround: Copy the
/var/lib/glusterd/vols/VOLNAME/infofile from existing node to new node and restart the glusterd on new node.
- BZ# 1201820When a snapshot is deleted, the corresponding file system object in the User Serviceable Snapshot is also deleted. Any subsequent file system access results in the snapshot Daemon becoming unresponsive.Workaround: Ensure that you do not perform any file system operations on the snapshot that is about to be deleted.
- BZ# 1191033When a large number of snapshots are activated and are accessed via User Serviceable Snapshots, the inode limit for the mounted filesystem may get exhausted, causing the deletion of inode entries from the inode table. Accessing the
.snapsdirectory when this happens may result in
ENOENTerrors in FUSE, and
Invalid Argumenterrors in NFS.Workaround: Clear the kernel VFS cache by executing the following command:
# echo 3 >/proc/sys/vm/drop_caches
- BZ# 1176835When the User Serviceable Snapshot feature is enabled, performing any operation or system call that uses the
statvfs()system calls on a
.snapsrepository fails with the following error:
Stale file handle
- BZ# 1160621If the current directory is not a part of the snapshot, for example,
snap1, then the user cannot enter the
- BZ# 1169790When a volume is down and there is an attempt to access
.snapsdirectory, a negative cache entry is created in the Kernal Virtual File System(VFS) cache for the
.snapsdirectory. After the volume is brought back online, accessing the
.snapsdirectory fails with an ENOENT error because of the negative cache entry.Workaround: Clear the kernel VFS cache by executing the following command:
# echo 3 >/proc/sys/vm/drop_caches
- BZ# 1170145After the restore operation is complete, if restore a volume while you are in the
.snapsdirectory, the following error message is displayed from the mount point -
"No such file or directory".Workaround:
- Navigate to the parent directory of the
- Drop VFS cache by executing the following command:
# echo 3 >/proc/sys/vm/drop_caches
- Change to the
- BZ# 1170365Virtual inode numbers are generated for all the files in the
.snapsdirectory. If there are hard links, they are assigned different inode numbers instead of the same inode number.
- BZ# 1170502On enabling the User Serviceable Snapshot feature, if a directory or a file by name
.snapsexists on a volume, it appears in the output of the
- BZ# 1174618If the User Serviceable Snapshot feature is enabled, and a directory has a pre-existing
.snapsfolder, then accessing that folder can lead to unexpected behavior.Workaround: Rename the pre-existing
.snapsfolder with another name.
- BZ# 1167648Performing operations which involve client graph changes such as volume set operations, restoring snapshot etc eventually leads to out of memory scenarios for the client processes which mount the volume.
- BZ# 1141433An incorrect output is displayed while setting the
snap-max-soft-limitoptions for volume or system/cluster.Example:
The same example is applicable for
- Command :
gluster snapshot config snap-max-hard-limit 100Output : snapshot config : System for snap-max-hard-limit set successfullyExpected Output : snapshot config : snap-max-hard-limit for the cluster set successfully.
- Command :
gluster snapshot config vol1 snap-max-hard-limit 100Output : snapshot config : vol1 for snap-max-hard-limit set successfully.Expected output : snapshot config : snap-max-hard-limit for vol1 set successfully.
- BZ# 1133861New snap bricks fails to start if the total snapshot brick count in a node goes beyond 1K.Workaround: Deactivate unused snapshots.
- BZ# 1126789If any node or
glusterdservice is down when snapshot is restored then any subsequent snapshot creation fails.Workaround: Do not restore a snapshot, if node or
glusterdservice is down.
- BZ# 1122064The snapshot volumes does not handshake based on versions. If a node or glusterd service is down when snapshot activate or deactivate command executed, the node on which the command was executed is not updated when the node or glusterd service is up and continues to be in the same state.
- BZ# 1139624While taking snapshot of a gluster volume, it creates another volume which is similar to the original volume. Gluster volume consumes some amount of memory when it is in started state, so as Snapshot volume. Hence, the system goes to out of memory state.Workaround: Deactivate unused snapshots to reduce the memory foot print.
- BZ# 1129675If glusterd is down in one of the nodes in cluster or if the node itself is down, then performing a snapshot restore operation leads to the inconsistencies:
Workaround: Perform snapshot restore only if all the nodes and their corresponding glusterd services are running.Restart glusterd service using
gluster volume heal vol-name infocommand displays the error message Transport endpoint not connected.
- Error occurs when clients try to connect to glusterd service.
# service glusterd startcommand.
- BZ# 1105543When a node with old snap entry is attached to the cluster, the old entries are propagated throughout the cluster and old snapshots which are not present are displayed.Workaround: Do not attach a peer with old snap entries.
- BZ# 1104191The Snapshot command fails if snapshot command is run simultaneously from multiple nodes when high write or read operation is happening on the origin or parent volume.Workaround: Avoid running multiple snapshot commands simultaneously from different nodes.
- BZ# 1059158The
NFS mountoption is not supported for snapshot volumes.
- BZ# 1114015A wrong warning message, Changing snapshot-max-hard-limit will lead to deletion of snapshots if they exceed the new limit. Do you want to continue? (y/n) is displayed while setting the configuration options (snap-max-hard-limit, snap-max-soft-limit) for system or volume. The snapshot will not be deleted even if configuration values are changed.
- BZ# 1113510The output of
gluster volume infoinformation (snap-max-hard-limit and snap-max-soft-limit) even though the values that are not set explicitly and must not be displayed.
- BZ# 1111479Attaching a new node to the cluster while snapshot delete was in progress, deleted snapshots successfully but gluster snapshot list shows some of the snaps are still present.Workaround: Do not attach or detach new node to the trusted storage pool operation while snapshot is in progress.
- BZ# 1092510If you create a snapshot when the rename of directory is in progress (here, its complete on hashed subvolume but not on all of the subvolumes), on snapshot restore, directory which was undergoing rename operation will have same GFID for both source and destination. Having same GFID is an inconsistency in DHT and can lead to undefined behavior.This is since in DHT, a rename (source, destination) of directories is done first on hashed-subvolume and if successful, then on rest of the subvolumes. At this point in time, if you have both source and destination directories present in the cluster with same GFID - destination on hashed-subvolume and source on rest of the subvolumes. A parallel lookup (on either source or destination) at this time can result in creation of directories on missing subvolumes - source directory entry on hashed and destination directory entry on rest of the subvolumes. Hence, there would be two directory entries - source and destination - having same GFID.
- BZ# 1104635During snapshot delete, if a node goes down, the snapshot delete command fails. Stale entries would be present in the node which went down and when the node comes back online, the stale entry is propagated to other nodes and this results an invalid snapshot entry which may not be deletable using CLI.Workaround:Manually delete the snapshot, including the back-end LVM, from all the nodes and restart
glusterdservice on all nodes.
- BZ# 1136207Volume status service shows All bricks are Up message even when some of the bricks are in unknown state due to unavailability of glusterd service.
- BZ# 1109683When a volume has a large number of files to heal, the
volume self heal infocommand takes time to return results and the nrpe plug-in times out as the default timeout is 10 seconds.Workaround:In
/etc/nagios/gluster/gluster-commands.cfgincrease the timeout of nrpe plug-in to 10 minutes by using the -t option in the command.Example:$USER1$/gluster/check_vol_server.py $ARG1$ $ARG2$ -o self-heal -t 600
- BZ# 1094765When certain commands invoked by Nagios plug-ins fail, irrelevant outputs are displayed as part of performance data.
- BZ# 1107605Executing
sadfcommand used by the Nagios plug-ins returns invalid output.Workaround: Delete the datafile located at
/var/log/sa/saDDwhere DD is current date. This deletes the datafile for current day and a new datafile is automatically created and which is usable by Nagios plug-in.
- BZ# 1107577The Volume self heal service returns a WARNING when there unsynchronized entries are present in the volume, even though these files may be synchronized during the next run of self-heal process if
self-healis turned on in the volume.
- BZ# 1109744Notification message is sent only once regarding quorum loss and not for each volume.
- BZ# 1121009In Nagios, CTDB service is created by default for all the gluster nodes regardless of whether CTDB is enabled on the Red Hat Storage node or not.
- BZ# 1089636In the Nagios GUI, incorrect status information is displayed as Cluster Status OK : None of the Volumes are in Critical State, when volumes are utilized beyond critical level.
- BZ# 1109739Cluster quorum monitoring Nagios plug-in displays only the latest messages received, so the user is unable to determine all the volumes for which quorum is lost.Workaround:Event log contains the list of the messages received. Also, if quorum is lost on a cluster, all volumes with
quorum-typeserver will be affected.
- BZ# 1079289When the memory utilization is very high, some or all the services may go to critical state and display the message: CHECK_NRPE: Socket timeout after 10 seconds.
- BZ# 1106421When cluster.quorum-type is set to none for all volumes in the cluster, the Cluster quorum monitoring plug-in only receives passive checks based on rsyslog messages and hence remains in Pending state as there are no service notifications available.
- BZ# 1085331Volume Utilization graph displays the error perfdata directory for host_directory for host_name does not exist, when volume utilization data is not available.
- BZ# 1111828In Nagios GUI, Volume Utilization graph displays an error when volume is restored using its snapshot.
- BZ# 1110282Executing
rebalance statuscommand, after stopping rebalance process, fails and displays a message that the rebalance process is not started.
- BZ# 1140531Extended attributes set on a file while it is being migrated during a rebalance operation are lost.Workaround: Reset the extended attributes on the file once the migration is complete.
- Rebalance does not proceed if any of the subvolumes of dht in the volume are down. This could be any brick in the case of a pure distributed volume. In a distributed replicated set, rebalance will proceed as long as at least one brick of each replica set is up.While running rebalance on a volume, ensure that all the bricks of the volume are in the operating or connected state.
- BZ# 960910After executing
rebalanceon a volume, running the
rm -rfcommand on the mount point to remove all of the content from the current working directory recursively without being prompted may return Directory not Empty error message.
- BZ# 862618After completion of the rebalance operation, there may be a mismatch in the failure counts reported by the
gluster volume rebalance statusoutput and the rebalance log files.
- BZ# 1039533While Rebalance is in progress, adding a brick to the cluster displays an error message,
failed to get indexin the gluster log file. This message can be safely ignored.
- BZ# 1064321When a node is brought online after rebalance, the status displays that the operation is completed, but the data is not rebalanced. The data on the node is not rebalanced in a remove-brick rebalance operation and running commit command can cause data loss.Workaround: Run the
rebalancecommand again if any node is brought down while rebalance is in progress, and also when the rebalance operation is performed after remove-brick operation.
- BZ# 1204044The script
stop-all-gluster-processes.shfails to stop the Geo-replication processes.Workaround:If the script fails on a Red Hat Storage node, manually search for the
gsyncdprocesses and kill them.
- BZ# 1179701If a master volume of a Red Hat Storage node goes into an irrecoverable state and it is replaced by a new node, the new node does not have changelogs because of which it performs a file system crawl operation and misses to sync a few files from the master volume.
- BZ# 1102524The Geo-replication worker goes to faulty state and restarts when resumed. It works as expected when it is restarted, but takes more time to synchronize compared to resume.
- BZ# 877895When one of the bricks in a replicate volume is offline, the
ls -lRcommand from the mount point reports Transport end point not connected.When one of the two bricks under replication goes down, the entries are created on the other brick. The Automatic File Replication translator remembers that the directory that is down contains stale data. If the brick that is online is killed before the self-heal happens on that directory, operations like readdir() fail.
- BZ# 1063830Performing add-brick or remove-brick operations on a volume having replica pairs when there are pending self-heals can cause potential data loss.Workaround: Ensure that all bricks of the volume are online and there are no pending self-heals. You can view the pending heal info using the command
gluster volume heal volname info.
- After the
gluster volume replace-brick VOLNAME Brick New-Brick commit forcecommand is executed, the file system operations on that particular volume, which are in transit, fail.
- After a replace-brick operation, the stat information is different on the NFS mount and the FUSE mount. This happens due to internal time stamp changes when the
replace-brickoperation is performed.
- BZ# 1001453Truncating a file to a larger size and writing to it violates the quota hard limit. This is since the XFS pre-allocation logic applied on the truncated file does not extract the actual disk space it consumed.
- BZ# 1003755Directory Quota feature does not work well with hard links. With a directory that has Quota limit set, the disk usage seen with the
du -hs directorycommand and the disk usage seen with the
gluster volume quota VOLNAME list directorycommand may differ. It is recommended that applications writing to a volume with directory quotas enabled, do not use hard links.
- BZ# 1016419Quota does not account for the disk blocks consumed by a directory. Even if a directory grows in size because of the creation of new directory entries, the size as accounted by quota does not change. You can create any number of empty files but you will not be able to write to the files once you reach the quota hard limit. For example, if the quota hard limit of a directory is 100 bytes and the disk space consumption is exactly equal to 100 bytes, you can create any number of empty files without exceeding quota limit.
- BZ# 1020275Creating files of different sizes leads to the violation of the quota hard limit.
- BZ# 1021466After setting Quota limit on a directory, creating sub directories and populating them with files and renaming the files subsequently while the I/O operation is in progress causes a quota limit violation.
- BZ# 998893Zero byte sized files are created when a write operation exceeds the available quota space. Since Quota does not account for the disk blocks consumed by a directory(as per Bug 1016419), the write operation creates the directory entry but the subsequent write operation fails because of unavailable disk space.
- BZ# 1023430When a quota directory reaches its limit, renaming an existing file on that directory leads to Quota violation. This is because the renamed is treated as a new file.
- BZ# 998791During a file rename operation if the hashing logic moves the target file to a different brick, then the rename operation fails if it is initiated by a non-root user.
- BZ# 999458Quota hard limit is violated for small quota sizes in the range of 10 MB to 100 MB.
- BZ# 1020713In a distribute or distribute replicate volume, while setting quota limit on a directory, if one or more bricks or one or more replica sets respectively, experience downtime, quota is not enforced on those bricks or replica sets, when they are back online. As a result, the disk usage exceeds the quota limit.Workaround: Set quota limit again after the brick is back online.
- BZ# 1032449In the case when two or more bricks experience a downtime and data is written to their replica bricks, invoking the quota list command on that multi-node cluster displays different outputs after the bricks are back online.
- Multi-client sequential write operations on the file system leads to the Red Hat Storage node to become unresponsive. See the Knowledge Centric Solution for more information on the steps to proceed further.
- After you restart the NFS server, the unlock within the grace-period feature may fail and the locks help previously may not be reclaimed.
- fcntl locking ( NFS Lock Manager) does not work over IPv6.
- You cannot perform NFS mount on a machine on which glusterfs-NFS process is already running unless you use the NFS mount
-o nolockoption. This is because glusterfs-nfs has already registered NLM port with portmapper.
- If the NFS client is behind a NAT (Network Address Translation) router or a firewall, the locking behavior is unpredictable. The current implementation of NLM assumes that Network Address Translation of the client's IP does not happen.
nfs.mount-udpoption is disabled by default. You must enable it to use posix-locks on Solaris when using NFS to mount on a Red Hat Storage volume.
- If you enable the
nfs.mount-udpoption, while mounting a subdirectory (exported using the
nfs.export-diroption) on Linux, you must mount using the
-o proto=tcpoption. UDP is not supported for subdirectory mounts on the GlusterFS-NFS server.
- For NFS Lock Manager to function properly, you must ensure that all of the servers and clients have resolvable hostnames. That is, servers must be able to resolve client names and clients must be able to resolve server hostnames.
- BZ# 1040418The length of the argument to
nfs.export-dir(or any other gluster set option) is limited to internal buffer size of the Linux shell. In a typical set up, the default size of this buffer is 131072 bytes.
- BZ# 1116374The nfs-ganesha daemon crashes if started in NIV_FULL_DEBUG level.Workaround:Use other log levels supported by nfs-ganesha while starting the server.
- BZ# 1116336The nfs-ganesha process is remains active after setting
kill -s TERMcommand does not kill nfs-ganesha.Workaround:Use kill -9 on the process ID of ganesha.nfsd process and then use CLI options to export new entries.
- BZ# 1115901Multi-node nfs-ganesha is not supported in this release.Workaround:In a multi-node volume setup, perform all CLI commands and steps on one of the nodes only.
- BZ# 1114574Executing
rpcinfo -pcommand after stopping nfs-ganesha displays NFS related programsWorkaround: Run
rpcinfo -dcommand each of the NFS related services listed in
rpcifnfo -pand start the Red Hat Storage volume forcefully using the following command:
# gluster vol start volume force
- BZ# 1091936When ACL support is enabled, getattr of ACL attribute on the files with no ACLs set return value as NULL. This leads to discrepancies while trying to read ACLs on the files present in the system.
- BZ# 1054124After files and directories are created on the mount point with root-squash enabled for nfs-ganesha, executing
nfsnobody:nfsnobody. This is because the client maps only 16 bit unsigned representation of -2 to nfsnobody whereas 4294967294 is 32 bit equivalent of -2. This is currently a limitation in upstream nfs-ganesha 2.0 and will be fixed in the future release.
- BZ# 1054739As multiple sockets are used with nfs-ganesha, executing
showmount -ecommand displays duplicate information.
- The GET and PUT commands fail on large files while using Unified File and Object Storage.Workaround: You must set the
node_timeout=60variable in the proxy, container, and the object server configuration files.
- BZ# 984591After stopping a Geo-replication session, if the files synced to the slave volume are renamed then when Geo-replication starts again, the renamed files are treated anew, (without considering the renaming) and synced on to the slave volumes again. For example, if 100 files were renamed, you would find 200 files on the slave side.
- BZ# 987929While the
rebalanceprocess is in progress, starting or stopping a Geo-replication session results in some files not get synced to the slave volumes. When a Geo-replication sync process is in progress, running the
rebalancecommand causes the Geo-replication sync process to stop. As a result, some files do not get synced to the slave volumes.
- BZ# 1029799Starting a Geo-replication session when there are tens of millions of files on the master volume takes a long time to observe the updates on the slave mount point.
- BZ# 1026072The Geo-replication feature keeps the status details including the
changelogentires in the
/var/run/glusterdirectory. On Red Hat Storage Server, this directory is a
tmpfsmountpoint, therefore there is a data loss after a reboot.
- BZ# 1027727When there are hundreds of thousands of hard links on the master volume prior to starting the Geo-replication session, some hard links are not getting synchronized to the slave volume.
- BZ# 1029899During a Geo-replication session, after you set the checkpoint, and subsequently when one of the active nodes goes down, the passive node replaces the active node. At this point the checkpoint for replaced node is displayed as invalid.
- BZ# 1030256During a Geo-replication session, when create and write operations are in progress, if one of the active nodes goes down, there is a possibility for some files to undergo a synchronization failure to the slave volume.
- BZ# 1063028When geo-replication session is running between master and slave, ACLs on the master volume are not reflected on the slave as ACLs (which are extended attributes) are not synced to the slave by Geo replication.
- BZ# 1056226User-set xattrs are not synced to the slave as Geo-replication does not process SETXATTR fops in changelog (and in hybrid crawl).
- BZ# 1063229After the upgrade, two Geo-replication monitor processes run for the same session. Both process try to use the same
xsync changelogfile to record the changes.Workaround: Before running
geo-rep create forcecommand, kill the Geo-replication monitor process.
- BZ# 877988Entry operations on replicated bricks may have a few issues with
md-cachemodule enabled on the volume graph.For example: When one brick is down, while the other is up an application is performing a hard link call
link()would experience EEXIST error.Workaround: Execute this command to avoid this issue:
# gluster volume set VOLNAME stat-prefetch off
- BZ# 986090Currently, the Red Hat Storage server has issues with mixed usage of hostnames, IPs and FQDNs to refer to a peer. If a peer has been probed using its hostname but IPs are used during add-brick, the operation may fail. It is recommended to use the same address for all the operations, that is, during peer probe, volume creation, and adding/removing bricks. It is preferable if the address is correctly resolvable to a FQDN.
- BZ# 882769When a volume is started, by default the NFS and Samba server processes are also started automatically. The simultaneous use of Samba or NFS protocols to access the same volume is not supported.Workaround: You must ensure that the volume is accessed either using Samba or NFS protocols.
- BZ# 852293The management daemon does not have a rollback mechanism to revert any action that may have succeeded on some nodes and failed on the those that do not have the brick's parent directory. For example, setting the
volume-idextended attribute may fail on some nodes and succeed on others. Because of this, the subsequent attempts to recreate the volume using the same bricks may fail with the error brickname or a prefix of it is already part of a volume.Workaround:
- You can either remove the brick directories or remove the glusterfs-related extended attributes.
- Try creating the volume again.
- BZ# 994950An input-output error is seen instead of the Disk quota exceeded error when the quota limit exceeds. This issue is fixed in the Red Hat Enterprise Linux 6.5 Kernel.
- BZ# 913364An NFS server reboot does not reclaim the file LOCK held by a Red Hat Enterprise Linux 5.9 client.
- BZ# 896314GlusterFS Native mount in Red Hat Enterprise Linux 5.x shows lower performance than the RHEL 6.x versions for high burst I/O applications. The FUSE kernel module on Red Hat Enterprise Linux 6.x has many enhancements for dynamic write page handling and special optimization for large burst of I/O.Workaround: It is recommended that you use Red Hat Enterprise Linux 6.x clients if you observe a performance degradation on the Red Hat Enterprise Linux 5.x clients.
- BZ# 1017728On setting the quota limit as a decimal digit and setting the
deem-statfson, a difference is noticed in the values displayed by the
df -hcommand and
gluster volume quota VOLNAME listcommand. In case of the
gluster volume quota VOLNAME listcommand, the values do not get rounded off to the next integer.
- BZ# 1030438On a volume, when read and write operations are in progress and simultaneously a rebalance operation is performed followed by a remove-brick operation on that volume, then the
rm -rfcommand fails on a few files.
- BZ# 1100590The
cp -aoperation from the NFS mount point hangs if barrier is already enabled.
- Mounting a volume with
-o aclcan negatively impact the directory read performance. Commands like recursive directory listing can be slower than normal.
- When POSIX ACLs are set and multiple NFS clients are used, there could be inconsistency in the way ACLs are applied due to attribute caching in NFS. For a consistent view of POSIX ACLs in a multiple client setup, use the -o noac option on the NFS mount to disable attribute caching. Note that disabling the attribute caching option could lead to a performance impact on the operations involving the attributes.
- BZ# 1013151Accessing a Samba share may fail if GlusterFS is updated while Samba is running.Workaround: On each node where GlusterFS is updated, restart Samba services after GlusterFS is updated.
- BZ# 994990When the same file is accessed concurrently by multiple users for reading and writing. The users trying to write to the same file will not be able to complete the write operation because of the lock not being available.Workaround: To avoid the issue, execute the command:
# gluster volume set VOLNAME storage.batch-fsync-delay-usec 0
- BZ# 1031783If Red Hat Storage volumes are exported by Samba, NT ACLs set on the folders by Microsoft Windows clients do not behave as expected.
server.event-threadsvalues to a lower value reduces the number of epoll threads only when subsequent IO operations are performed on the mount point of the volume.
sosreportcommand does not collect the gluster command logs without flags or options in the
sosreport -acommand to collect logs in the
- If files and directories have different GFIDs on different back-ends, the glusterFS client may hang or display errors.Contact Red Hat Support for more information on this issue.
- BZ# 920002The POSIX compliance tests fail in certain cases on Red Hat Enterprise Linux 5.9 due to mismatched timestamps on FUSE mounts. These tests pass on all the other Red Hat Enterprise Linux 5.x and Red Hat Enterprise Linux 6.x clients.
- BZ# 916834The
quick-readtranslator returns stale file handles for certain patterns of file access. When running the dbench application on the mount point, a dbench: read fails on handle 10030 message is displayed.Work Around: Use the command below to avoid the issue:
# gluster volume set VOLNAME quick-read off
- BZ# 1030962On installing the Red Hat Storage Server from an ISO or PXE, the
kexec-toolspackage for the
kdumpservice gets installed by default. However, the
crashkernel=autokernel parameter required for reserving memory for the
kdumpkernel, is not set for the current kernel entry in the bootloader configuration file,
/boot/grub/grub.conf. Therefore the
kdumpservice fails to start up with the following message available in the logs.
kdump: No crashkernel parameter specified for running kernelWorkaround: After installing the Red Hat Storage Server, the
crashkernel=auto, or an appropriate
crashkernel=sizeMkernel parameter can be set manually for the current kernel in the bootloader configuration file. After that, the Red Hat Storage Server system must be rebooted, upon which the memory for the
kdumpkernel is reserved and the
kdumpservice starts successfully. Refer to the following link for more information on Configuring kdump on the Command LineAdditional information: On installing a new kernel after installing the Red Hat Storage Server, the
crashkernel=autokernel parameter is successfully set in the bootloader configuration file for the newly added kernel.
- BZ# 866859The sosreport behavior change (to glusterfs and sosreport) is altered in the statedump behavior configuration file (
glusterdump.optionsfile) and it is placed in /tmp. This file has information on the path and options you can set on the behavior of the
statedumpfile. The glusterfs daemon searches for this file and subsequently places the
statedumpinformation in the specified location. Workaround: Change the configurations in glusterfs daemon to make it look at /usr/local/var/run/gluster for
glusterdump.optionsfile by default. No changes to be performed to make sosreport write its configuration file in /usr/local/var/run/gluster instead of /tmp.
- BZ# 1054759A vdsm-tool crash report is detected by Automatic Bug Reporting Tool (ABRT) in Red Hat Storage Node as the
/etc/vdsm/vdsm.idfile was not found during the first time.Workaround: Execute the command
/usr/sbin/dmidecode -s system-uuid > /etc/vdsm/vdsm.idbefore adding the node to avoid the vdsm-tool crash report.
- BZ# 1058032While migrating VMs, libvirt changes the ownership of the guest image, unless it detects that the image is on a shared filesystem and the VMs can not access the disk imanges as the required ownership is not available.Workaround: Perform the steps:
- Power-off the VMs before migration.
- After migration is complete, restore the ownership of the VM Disk Image (107:107)
- Start the VMs after migration.
- BZ# 990108Volume options that begin with user.* are considered user options and these options cannot be reset as there is no way of knowing the default value.
- BZ# 1065070The python-urllib3 package fails to downgrade and this in turn results in Red Hat Storage downgrade process failure.Workaround: Move the
/tmpand perform a fresh installation of the python-urllib3 package.
- BZ# 1086159The glusterd service crashes when volume management commands are executed concurrently with peer commands.
- BZ# 1130270If a 32 bit Samba package is installed before installing Red Hat Storage Samba package, the installation fails as Samba packages built for Red Hat Storage do not have 32 bit variantsWorkaround:Uninstall 32 bit variants of Samba packages.
- BZ# 1139183The Red Hat Storage 3.0 version does not prevent clients with versions older Red Hat Storage 3.0 from mounting a volume on which rebalance is performed. Users with versions older than Red Hat Storage 3.0 mounting a volume on which rebalance is performed can lead to data loss.You must install latest client version to avoid this issue.
- BZ# 1127178If a replica brick goes down and comes up when
rm -rfcommand is executed, the operation may fail with the message Directory not empty.Workaround: Retry the operation when there are no pending self-heals.
- BZ# 969020Renaming a file during remove-brick operation may cause the file not to get migrated from the removed brick.Workaround: Check the removed brick for any files that might not have been migrated and copy those to the gluster volume before decommissioning the brick.
- BZ# 1007773When
remove-brick startcommand is executed, even though the graph change is propagated to the NFS server, the directory inodes in memory are not refreshed to exclude the removed brick. Hence, new files that are created may end up on the removed-brick.Workaround: If files are found on the removed-brick path after
remove-brick commit, copy them via a gluster mount point before re-purposing the removed brick.
- BZ# 1116115When I/O operation is being performed at a high rate on the volume, faster than the rate at which the quota's accounting updates disk usage, the disk usage crosses the soft-limit mark without being failed with EDQUOT.Workaround: Set
features.hard-timeoutto a lower value, depending on the workload. Setting
features.hard-timeoutto zero ensures that the disk usage accounting happens in line with the I/O operations performed on the volume, depending on the applicable timeout, but it could result in an increase in I/O latencies.
features.soft-timeoutapplies when the disk usage is lesser than the quota soft-limit set on a directory.
features.hard-timoutapplies when the disk usage is greater than the quota soft-limit but lesser than the hard-limit set on a directory.
- BZ# 1116121When I/O is being performed at a high rate on the volume, faster than the rate at which the quota's accounting updates disk usage, the disk usage crosses the soft-limit mark without an alert being logged.Workaround: Set
features.soft-timeoutto a lower value, depending on the workload. Setting
features.soft-timeoutto zero would ensure that the disk usage accounting happens in line with the I/O performed on the volume, but it could result in an increase in I/O latencies.
- BZ# 1120437Executing
peer-statuscommand on probed host displays the IP address of the node on which the peer probe was performed.Example When probing a peer, node B with hostname from node A, executing
peer statuscommand on node B, displays IP address of node A instead of its hostname.Workaround: Probe node A from node B with hostname of node A. For example, execute the command:
# gluster peer probe HostnameAfrom node B.
- BZ# 1097309If a NFS-client gets disconnected from the Red Hat Storage server without releasing the locks it obtained, these locks can prevent other NFS clients or Native (FUSE) clients to access the locked files. The NFSv3 protocol does not allow releasing the locks server side without a restart of the NFS servers. A restart triggers a grace period where existing locks need to get re-obtained. Locks are expired and made available to other NFS clients when the previous NFS clients do not re-request the previously obtained locks. More details related to NFS clients and lock recovery can be found in https://access.redhat.com/site/solutions/895263.
- BZ# 1122371The NFS server process and gluster self-heal daemon process restarts when gluster daemon process is restarted.
- BZ# 1110692Executing
remove-brick statuscommand, after stopping remove-brick process, fails and displays a message that the remove-brick process is not started.
- BZ# 1123733Executing a command which involves glusterd-glusterd communication Example
gluster volume statusimmediately after one of the nodes is down hangs and fails after 2 minutes with cli-timeout message. The subsequent command fails with the error message Another transaction in progress for 10 mins (frame timeout).Workaround: Set a non-zero value for ping-timeout in
- BZ# 1115915When a GlusterFS-native (FUSE) client looses its connection to the storage server without properly closing it, the brick process does not release resources (like locks) within an acceptable time. Other GlusterFS-native clients that require these resources can get blocked until the TCP-connection gets garbage collected by networking layers in the kernel and the brick processes get informed about it. This can introduce delays of 15-20 minutes before locks are released.Workaround: Reduce the value of the system-wide
net.ipv4.tcp_retries2 sysctl. Due to this change, the network layer of the Linux kernel times-out TCP-connections sooner.
- BZ# 1136718The afr self-heal can leave behind a partially healed file if the brick containing afr self-heal source file goes down in the middle of heal operation. If this partially healed file is migrated before the brick that was down comes online again, the migrated file would have incorrect data and the original file would be deleted.
- BZ# 1139193After
add-brickoperation, any application (like git) which attempts
opendiron a previously present directory fails with
- BZ# 1141172If you rename a file from multiple mount points, there are chances of losing the file. This issue is witnessed since
mvcommand sends unlinks instead of renames when source and destination happens to be hard links to each other. Hence, the issue is in mv, distributed as part of
coreutilsin various Linux distributions.For example, if there are parallel renames of the form (mv a b) and (mv b a) where a and b are hard links to the same file, because of the above mentioned behavior of mv, unlink (a) and unlink (b) would be issued from both instances of
mv. This results in losing both the links a and b and hence the file.
- BZ# 979926When any process establishes a TCP connection with
glusterfsservers of a volume using port
> 1023, the server rejects the requests and the corresponding file or management operations fail. By default,
glusterfsservers treat ports
> 1023as unprivileged.Workaround: To disable this behavior, enable
rpc-auth-allow-insecureoption on the volume using the steps given below:
- To allow
insecureconnections to a volume, run the following command:
#gluster volume set VOLNAME rpc-auth-allow-insecure on
- To allow
insecureconnections to glusterd process, add the following line in
option rpc-auth-allow-insecure on
glusterdprocess using the following command:
# service glusterd restart
- Restrict connections to trusted clients using the following command:
#gluster volume set VOLNAME auth.allow IP address
- BZ# 1139676Renaming a directory may cause both source and target directories to exist on the volume with the same GFID and make some files in these directories not visible from the mount point. The files will still be present on the bricks.Workaround: The steps to fix this issue are documented in: https://access.redhat.com/solutions/1211133
- BZ# 1030309During directory creations attempted by geo-replication, though an
EEXIST, the directory might not have a complete layout for sometime and the directory creation fails with
Directory existsmessage. This can happen if there is a parallel
mkdirattempt on the same name. Till the other
mkdircompletes, layout is not set on the directory. Without a layout, entry creations within that directory fails.Workaround: Set the layout on those subvolumes where the directory is already created by the parallel
mkdirbefore failing the current
NoteThis is not a complete fix as the other
mkdirmight not have created directories on all subvolumes. The layout is set on the subvolumes where directory is already created. Any file or directory names which hash to these subvolumes on which layout is set, can be created successfully.
3.2. Red Hat Storage Console
- BZ#1164662The Trends tab in the Red Hat Storage Console appears to be empty after the ovirt engine restarts. This is due to the Red Hat Storage Console UI-plugin failing to load on the first instance of restarting the ovirt engine.Workaround: Refresh(F5) the browser page to load the Trends tab.
- BZ#1165269When you add a Red Hat Storage node in the Red Hat Storage Console using its IP address and remove it from the Red Hat Storage trusted Storage Pool, and consequently use the FQDN of the node to add it again to the Trusted Storage pool, the operation fails. This results in the node becoming unresponsive.Workaround: Remove the unresponsive node from the Red Hat Storage Trusted Storage Pool and try adding it again.
- BZ#1167305The Trends tab on the Red Hat Storage Console does not display the thin-pool utilization graphs in addition to the brick utilization graphs. Currently, there is no mechanism for the UI plugin to detect if the volume is provisioned using the thin provisioning feature.
- BZ#1167572On editing the cluster version in the Edit Cluster dialog box on the Red Hat Storage Console, the compatible version field gets loaded with the highest available compatibility version by default, instead of the current version of the cluster.Workaround: Select the correct version of the cluster in the Edit Cluster dialog box before clicking on the OK button.
- BZ#916095If Red Hat Storage node is added to the cluster using IP address and the same Red Hat Storage node is later added using the FQDN (Fully Qualified Domain Name), the installation fails.
- BZ# 990108Resetting the
user.cifsoption using the Create Volume operation on the Volume Options tab on the Red Hat Storage Console reports a failure.
- BZ# 978927Log messages that Red Hat Storage Console is trying to update VM/Template information are displayed.
- BZ# 880509When run on versions higher than Firefox 17, the Red Hat Storage Console login page displays a browser incompatibility warning. Red Hat Storage Console can be best viewed in Firefox 17 and higher versions.
- BZ# 1049759When
rhsc-log-collectorcommand is run, after collecting logs from different servers, the Terminal becomes garbled and unusable.Workaround: Run the
- BZ# 1054366In Internet Explorer 10, while creating a new cluster with Compatibility version 3.3, the Host drop down list does not open correctly. Also, if there is only one item, the drop down list gets hidden when the user clicks on it.
- BZ# 1053395In Internet Explorer, while performing a task, an error message Unable to evaluate payload is displayed.
- BZ# 1056372When no migration is occurring, incorrect error message is displayed for the stop migrate operation.
- BZ# 1049890When gluster daemon service is restarted, failed Rebalance is started automatically and the status is displayed as Started in the Red Hat Storage Console.
- BZ# 1048426When there are more entries in Rebalance Status and remove-brick Status window, the column names scrolls up along with the entries while scrolling the window.Workaround: Scroll up the Rebalance Status and remove-brick Status window to view the column names.
- BZ# 1053112When large sized files are migrated, the stop migrate task does not stop the migration immediately but only after the migration is complete.
- BZ# 1040310If the Rebalance Status dialog box is open in the Red Hat Storage Console while Rebalance is being stopped from the Command Line Interface, the status is currently updated as Stopped. But if the Rebalance Status dialog box is not open, the task status is displayed as Unknown because the status update relies on the gluster Command Line Interface.
- BZ# 1051696When a cluster with compatibility version 3.2 contains Red Hat Storage 2.1 U2 nodes, creating Volume with bricks in root partition fails and the force option to allow bricks in root partition is not displayed.Workaround: Do not create bricks in root partition or move the Cluster Compatibility Version to 3.3.
- BZ# 838329When incorrect create request is sent through REST api, an error message is displayed which contains the internal package structure.
- BZ# 1049863When Rebalance is running on multiple volumes, viewing the brick advanced details fails and the error message could not fetch brick details, please try again later is displayed in the Brick Advanced Details dialog box.
- BZ# 1022955Rebalance or remove-brick cannot be started immediately after stopping Rebalance or remove-brick, when a large file migration is in progress, as part of the previous operation (rebalance or remove-brick), even though it says it has stopped.
- BZ# 1015455The information on successfully completed Rebalance volume task is cleared from the Red Hat Storage Console after 5 minutes. The information on failed tasks is cleared after 1 hour.
- BZ# 1038691The RESTful Service Description Language (RSDL) file displays only the response type and not the detailed view of the response elements.Workaround: Refer the URL/API schema for detailed view of the elements of response type for the actions.
- BZ# 1024184If there is an error while adding bricks, all the "." characters of FQDN / IP address in the error message will be replaced with "_" characters.
- BZ# 982625Red Hat Storage Console allows adding Red Hat Storage 2.0+ and Red Hat Storage 2.1 servers into a 3.0 Cluster which is not supported in Red Hat Storage.
- BZ# 975399When Gluster daemon service is restarted, the host status does not change to UP from Non-Operational immediately in the Red Hat Storage Console. There would be a 5 minute interval for auto-recovery operations which detect changes in Non-Operational hosts.
- BZ# 971676While enabling or disabling Gluster hooks, the error message displayed if all the servers are not in UP state is incorrect.
- BZ# 1054759A
vdsm-tool crashreport is detected by Automatic Bug Reporting Tool (ABRT) in Red Hat Storage Node as the
/etc/vdsm/vdsm.idfile was not found during the first time.Workaround: Execute the command
/usr/sbin/dmidecode -s system-uuid > /etc/vdsm/vdsm.idbefore adding the node to avoid the
- BZ# 1057122While configuring the Red Hat Storage Console to use a remote database server, on providing either
noas input for
Database host name validationparameter, it is considered as
- BZ# 1042808When remove-brick operation fails on a volume, the Red Hat Storage node does not allow any other operation on that volume.Workaround: Perform commit or stop for the failed remove-brick task, before another task can be started on the volume.
- BZ# 1060991In Red Hat Storage Console, Technology Preview warning is not displayed for stop remove-brick operation.
- BZ# 1057450Brick operations like adding and removing a brick from Red Hat Storage Console fails when Red Hat Storage nodes in the cluster have multiple FQDNs (Fully Qualified Domain Names).Workaround: Host with multiple interfaces should map to the same FQDN for both Red Hat Storage Console and gluster peer probe.
- BZ# 958803When a brick process goes down, the brick status is not updated and displayed immediately in the Red Hat Storage Console as the Red Hat Storage Console synchronizes with the gluster Command Line Interface every 5 minutes for brick status.
- BZ# 1038663Framework restricts displaying delete actions for collections in RSDL display.
- BZ# 1061677When Red Hat Storage Console detects a remove-brick operation which is started from gluster Command Line Interface, engine does not acquire lock on the volume and Rebalance task is allowed.Workaround: Perform commit or stop on remove-brick operation before starting Rebalance.
- BZ# 1061813After stopping, committing, or retaining bricks from the Red Hat Storage Console UI, the details of files scanned, moved, and failed are not displayed in the Tasks pane.Workaround: Use Status option in Activities column to view the details of the remove-brick operation.
- BZ# 924826In Red Hat Storage Console, parameters related to Red Hat Enterprise Virtualization are displayed while searching for Hosts using the Search bar.
- BZ# 1062612When Red Hat Storage 2.1 Update 2 nodes are added to 3.2 cluster, users are allowed to perform Rebalance and remove-brick tasks which are not supported for 3.2 cluster.
- BZ# 977355When resolving a missing hook conflict, if one of the servers in the cluster is not online, an error message is displayed without the server name. Hence, the server which was down can not be identified.Workaround: Identify the information on the server which was down from the Hosts tab.
- BZ# 1046055While creating volume, if the bricks are added in root partition, the error message displayed does not contain the information that
Allow bricks in root partition and re-use the bricks by clearing xattrsoption needs to be selected to add bricks in root partition.Workaround: Select
Allow bricks in root partition and re-use the bricks by clearing xattrsoption to add bricks in root partition.
- BZ# 1060991In Red Hat Storage Console UI, Technology Preview warning is not displayed for stop remove-brick operation.
- BZ# 1066130Simultaneous start of Rebalance on volumes that span same set of hosts fails as gluster daemon lock is acquired on participating hosts.Workaround: Start Rebalance again on the other volume after the process starts on first volume.
- BZ# 1086718The Red Hat Access Plug-in related answers are not written to the answer file correctly when the
rhsc-setup --generate-answer=answer-filecommand is executed and hence the next offline execution of
rhsc-setup --offline --config-append=answer-file) asks Red Hat Access Plug-in related questions again.Workaround:
- Add the entry given below in the answer-file
- Execute the Red Hat Storage Console setup script in the offline mode
rhsc-setup --offline --config-append=answer-file
- BZ# 1086723Auto installation of Red Hat Storage Console using the answer file in Red Hat Storage Console-2.1.2 fails with the following warning message:[WARNING] Current installation does not support upgrade. It is recommended to clean your installation using rhsc-cleanup and re-install.Workaround:
- Execute the command:
yum update rhsc-setup
- Perform the Red Hat Storage Console auto installation using the
answer-filewith the command:
- BZ# 1201740The Red Hat Storage Console overrides the Red Hat Enterprise Linux values for the
dirty_background_ratioto 5 and 2 respectively. This occurs when activating the tuned profile
rhs-virtualizationwhile adding Red Hat Storage nodes to the Red Hat Storage Console. This could decrease the performance of the Red Hat Storage Trusted Storage Pool.Workaround:Restore the values of vm.dirty_ratio and dirty_background_ratio to the default values by activating the default tuned profile by executing the following command:
# tuned-adm profile default
- BZ# 1202731When you upgrade the Red Hat Storage Console from version 2.1.6 to 3.0.4, the
rhsc-log-collectorpackage does not get upgraded automatically.Workaround:Execute the following command to manually upgrade the package:
# yum upgrade rhsc-log-collector
- BZ# 1200248The Trends tab on the Red Hat Storage Console does not display all the network interfaces available on a host. This limitation is because the Red Hat Storage Console
ui-plugindoes not have this information.Workaround:The graphs associated with the hosts are available in the Nagios UI on the Red Hat Storage Console.You can view the graphs by clicking the Nagios home link
- BZ# 1194150The volumes of RDMA type are not supported by the Red Hat Storage Console. Such volumes can be created using the gluster CLI and the volumes are displayed on the UI. If the RDMA volume does not support TCP then TCP port is displayed as N/A and the corresponding brick appears offline. The bricks that belong to pure RDMA volumes which do not support TCP, appear offline on the console but work properly in the Red Hat Storage server.
- BZ# 1196144A few changes in the nrpe.cfg file requires restarting the nrpe service when the gluster-nagios-addons package is updated. As a result, the brick and disk monitoring does not work properly unless the nrpe service is restarted.Workaround:Restart the nrpe service using the following command:
# service nrpe reload
- BZ# 928926When you create a cluster through API, enabling both gluster_service and virt_service is allowed though this is not supported.
- BZ# 1059806In Red Hat Storage Console Command Line Interface, removing a cluster using its name fails with an error message and the same cluster gets deleted if UUID is used in remove command.
- BZ# 1108688An image in the Nagios home page is not transferred via SSL and the Security details displays the following message:Connection Partially Encrypted.
- BZ# 1111549In the Red Hat Storage Console, if the name provided in the
Namefield of the host in the New Host" pop-up is different from the Hostname provided in Nagios, the utilization details for the hosts are not displayed in Trends tab.
- BZ# 1100960Nagios does not support SELinux fully and this impacts rhsc-setup and normal usage of Red Hat Console with Nagios.Workaround: Run SELinux in permissive mode.
- BZ# 1113103The Network Utilization graph shows an error that RRD data does not exist.Workaround: After midnight of the next day, the sadf output file gets corrected automatically and graph works fine.
- BZ# 1121612The mount points for a host can not be detected from Trends tab in Red Hat Storage Console GUI.View mount point Disk Utilization using Nagios GUI.
- BZ# 1101181After stopping the rebalance/remove-brick operation from Red Hat Storage Console GUI, clicking on rebalance/remove-brick status button throws Unable to fetch data error.
- BZ# 1125960Both Red Hat Storage Console and Hadoop Ambari are using Nagios to monitor Red Hat Storage Server nodes, but there is version conflict for the package nagios-plugins, which blocks the installation of nagios required by Hadoop Ambari. Red Hat Stroage server 3.0 nodes are pre-bundled with latest version of nagios-plugin. If you want to use Red Hat Storage Console, then you will not be able to install Nagios via Ambari and utilize the pre-build hadoop monitors/alerts. HDP 2.0.6 is not supported for Red Hat Storage 3.0 release.
3.3. Red Hat Storage and Red Hat Enterprise Virtualization Integration
- In the case that the Red Hat Storage server nodes and the Red Hat Enterprise Virtualization Hypervisors are present in the same data center, the servers of both types are listed for selection when you create a virtual machine or add a storage domain. Red Hat recommends that you create a separate data center for the Red Hat Storage server nodes.
- BZ# 867236While deleting a virtual machine using the Red Hat Enterprise Virtualization Manager, the virtual machine is deleted but remains in the actual storage. This consumes unnecessary storage.Workaround: Delete the virtual machine manually using the command line interface. Deleting the virtual image file frees the space.
- BZ# 918032In this release, the
direct-io-mode=enablemount option does not work on the Hypervisor.
- BZ# 979901Virtual machines may experience very slow performance when a rebalance operation is initiated on the storage volume. This scenario is observed when the load on storage servers are extremely high. Hence, it is recommended to run the rebalance operation when the load is low.
- BZ# 856121When a volume starts, a
.glusterfsdirectory is created in the back-end export directory. When a
remove-brickcommand is performed, it only changes the volume configuration to remove the brick and stale data is present in back-end export directory.Workaround: Run this command on the Red Hat Storage Server node to delete the stale data.
# rm -rf /export-dir
- BZ# 866908The
gluster volume heal VOLNAME infocommand gives stale entries in its output in a few scenarios.
Workaround: Execute the command after 10 minutes. This removes the entries from internal data structures and the command does not display stale entries.
3.4. Red Hat Storage and Red Hat OpenStack Integration
- BZ# 1004745If a replica pair is down while taking a snapshot of a Nova instance on top of a Cinder volume hosted on a Red Hat Storage volume, the snapshot process may not complete as expected.
- BZ# 991490Mount options specified in
glusterfs_shares_configfile are not honored when it is specified as part of
- If storage becomes unavailable, the volume actions fail with
gluster volume delete VOLNAME forceto forcefully delete the volume.
- BZ# 1042801Cinder volume migration fails to migrate from one glusterFS back-end cluster to another. The migration fails even though the target volume is created.
- BZ# 1062848When a nova instance is rebooted while rebalance is in progress on the Red Hat Storage volume, the root file system will be mounted as read-only after the instance comes back up. Corruption messages are also seen on the instance.
Chapter 4. Technology Previews
4.1. gstatus Command
gstatuscommand provides an easy-to-use, high-level view of the health of a trusted storage pool with a single command. It gathers information by executing the GlusterFS commands, to gather information about the statuses of the Red Hat Storage nodes, volumes, and bricks.
4.2. Striped Volumes
4.3. Distributed-Striped Volumes
4.4. Distributed-Striped-Replicated Volumes
4.5. Striped-Replicated Volumes
4.6. Replicated Volumes with Replica Count greater than 3
4.7. Stop Remove Brick Operation
remove-brick stopcommand. The files that are already migrated during remove-brick operation, will not be reverse migrated to the original brick.
4.8. Read-only Volume
4.9. NFS Ganesha
4.10. Non Uniform File Allocation
Appendix A. Revision History
|Revision 3-10||Mon Jun 08 2015||Divya Muntimadugu|
|Revision 3-9||Wed Apr 08 2015||Pavithra Srinivasan|
|Revision 3-8||Mon Mar 30 2015||Divya Muntimadugu|
|Revision 3-7||Wed Mar 25 2015||Pavithra Srinivasan|