Red Hat Gluster Storage 3.5 update 5 Resolved Issues and Enhancements

Red Hat Gluster Storage 3.5 update 5 is a batch release for Red Hat Gluster Storage 3.5 and can be applied as an update to an existing installation or can be used to create a new installation. Red Hat Gluster Storage 3.5 update 5 updates are cumulative and include fixes from previous releases.

The following list provides errata for Red Hat Gluster Storage 3.5 update 5 and contains the complete list of fixes :

• samba bug fix and enhancement update
• samba bug fix and enhancement update
• nfs-ganesha bug fix and enhancement update
• nfs-ganesha bug fix and enhancement update
• glusterfs bug fix and enhancement update
• glusterfs bug fix and enhancement update

Notable Bug Fixes

• [1918018] nfs-ganesha does not start due to shared storage not being ready but returns 0 as success

  • Previously, nfs-ganesha failed with no indication to the user if the gluster_shared_storage volume did not mount while attempting to start nfs-ganesha. With this update, additional logic is added to the ganesha HA setup script to check if the gluster_shared_storage volume is mounted, and it fails with an error if not mounted. Now, gluster CLI reports that the startup of nfs-ganesha has failed and the user can take the necessary steps to resolve the issue.

• [1689375] Inconsistent locking with Tibco EMS

  • Previously, when a client got disconnected and reconnected, the client reopened the saved File Descriptors(FD) with subsequent operations. This led to problems in the applications where POSIX locks on FD’s were mandatory. With this update, a new option client.strict-locks is added, which when enabled prevents the reopening of saved FD’s. This option sets OFF by default, update all servers and clients to RHGS-3.5.5 and the cluster.op-version to 70200.

• This content is not included.1943467 Automatic File Replication(AFR) does not release eager locks for fsync when there is a contention

  • Previously in AFR, the fsync FOP did not have the logic of releasing the active client's eager-lock on the inode whenever there was a conflicting request from another client. As a result, no call-bailing of FOPS due to the blocked locks on RHHI-V workloads which led to a paused VM. With this update, it is fixed by making AFR fsync aware of conflicting locks so that it releases the active lock if there is a contention.

• This content is not included.1927640 Option to generate core dump without killing the process

  • With this update, the option to generate a core dump without killing the process is introduced.

• This content is not included.1992331 [rhel8] SIGSEGV crash after upgrade to RHGS 3.5.4

  • Previously, when a program tried to dereference a null pointer an uncommon condition occurred. It received multiple lock requests where some requests were blocked as the client info from the NLM subsys that required building a lease_id was not present. With this update, the client information from the NLM subsys is provided where possible.

• [1981440] [rhel8] nfs-ganesha: ganesha.nfsd killed by SIGSEGV

  • Previously, ganesha.nfsd crashed on invalid locks whenever a file was locked incorrectly. A file can be a directory or a link. As a result, the lock manager was not able to bail out with an error and ultimately crashed ganesha.nfsd. With this update, this problem is fixed by revising logic to return an error.

• [1981441] [rhel8] nfs-ganesha-3.4-2 ganesha.nfsd killed by SIGABRT

  • Previously, the ganesha.nfsd process doubled the freed memory after the client sent invalid permissions for a newly created file. With this update, it is fixed by revising logic which ensures that the memory is freed only once.

• [1925589] [rhel7] nfs-ganesha: ganesha.nfsd killed by SIGSEGV

  • Previously, ganesha.nfsd crashed on invalid locks whenever a file was locked incorrectly. A file can be a directory or a link. As a result, the lock manager was not able to bail out with an error and ultimately crashed ganesha.nfsd. With this update, this problem is fixed by revising logic to return an error.

• This content is not included.1995674 [rhel7] SIGSEGV crash after upgrade to RHGS 3.5.4

  • Previously, when a program tried to dereference a null pointer an uncommon condition occurred. It received multiple lock requests where some requests were blocked as the client info from the NLM subsys that required building a lease_id was not present. With this update, the client information from the NLM subsys is provided where possible.

• [1977267] [rhel7] nfs-ganesha-3.4-2 ganesha.nfsd killed by SIGABRT

  • Previously, the ganesha.nfsd process doubled the freed memory after the client sent invalid permissions for a newly created file. With this update, it is fixed by revising the logic which ensures that the memory is freed only once.

Support Limits

• For RHEL based Red Hat Gluster Storage upgrading to 3.5 batch update 5 with Samba, the write-behind translator has to be manually disabled for all existing samba volumes.

 # gluster volume set <volname> performance.write-behind disable

• The following features are considered deprecated in Red Hat Gluster Storage 3.5. Red Hat does not support it on new deployments and existing deployments that upgrade to Red Hat Gluster Storage 3.5 update 4.

  • Tiering
  • Gluster-NFS
    • Gluster NFS users are advised to migrate to NFS-Ganesha before upgrading to RHGS 3.5.3. NFS-Ganesha provides enhanced functionality, additional security features, and performance improvements.
  • Quota
  • SMB Multi-Channel IO (Tech Preview)

Known Issue

• Setting the auth.allow option can crash Gluster mounts in hundreds of client configurations if the setting dynamic-auth is not disabled.
  • Users who have configured hundreds of auth.allow addresses can encounter issues with Gluster mounts crashing while a client is attempting to mount or remount the volume. This is because of setvolume requests not completed as the dns-server is taking an extended time to resolve the address. Currently, we are attempting to resolve the address by sequentially going through configured hosts in the auth-list and then comparing it to the client address. If a client hostname is near the end of the auth-list then we can breach the timeout threshold (42 seconds) and crash the mount.
    For more information, contact Red Hat Support.

For RHGS deprecations

User's requiring WebAdmin should remain on RHEL 7, as WebAdmin is not supported on RHEL 8.

Bugs Fixed in Red Hat Gluster Storage 3.5 update 5