Oracle HSM 6.1 What Works With What
April 22, 2020
StorageTek Storage Archive Manager (SAM-QFS) was renamed to Oracle Hierarchical Storage Manager (Oracle HSM) at the 6.0 level.
Oracle HSM is an integrated hierarchical storage manager (HSM) and storage area network (SAN) shared file system (QFS). Oracle HSM also has integrated disk volume management and tape volume management. It also has a write once, read many times (WORM) file system capability. QFS can be used independently of Oracle HSM when just a file system is needed. Oracle HSM requires QFS and cannot be used independently of QFS. This "What Works With What" covers the specifics of what works with Oracle HSM.
This What Works With What (WWWW) covers the following products:
- Oracle HSM 6.1, and all of its patches
- StorageTek QFS 6.1, and all of its patches
- NOTE: The original Patch 6.1-02 has been recalled and replaced with Patch 6.1-03. This replacement patch is available at the Oracle Software Delivery Cloud (edelivery.oracle.com) 6.1-02 location.
The following products, configurations, and features are supported with Oracle HSM and QFS.
- All Sun SPARC (64 bit) architecture servers and workstations.
- All Sun x64 (64 bit) AMD architecture servers and workstations.
- All Sun x64 (64 bit) Intel architecture servers and workstations.
- All Sun x86 (32 bit) Intel architecture servers are only supported as Shared QFS clients.
- All Fujitsu Prime Power SPARC servers.
- All Sun SCSI, FC, and iSCSI protocols.
- All Sun SCSI and FC HBAs.
- All Sun FC switches.
- All Sun RAID disk arrays (includes StorageTek RAID disk arrays).
- All Sun hard disk drives (includes the X4500).
- NVMe SSD drives that require the use of the "blkdev" device driver are not supported currently.
- The tape libraries and tape drives listed at:
SAM-QFS Tape Library and Drive Support, 5.4 & beyond
Note that several old tape drives and libraries have been dropped from the support list at 5.4. This was announced previously in this page.
- All non-Sun HBAs and switches supported by Sun (includes HBAs and switches listed in the Interop and PreQual Tools as well as via the Get To Yes program).
- All non-Sun RAID disk arrays supported by Sun (includes RAID disk arrays listed in the Interop and PreQual Tools as well as via the Get To Yes program).
- Oracle Solaris Operating Systems: See section 4.1 below for certified versions
- Linux Operating Systems: See section 4.2 below for certified versions
- ZFS Volume Manager (ZVOLs)
- Solaris Volume Manager (SVM)
- Solaris Volume Manager (SVM) OBAN cluster capability with Solaris Cluster
- Veritas Volume Manager (VxVM) with stand-alone SAM-QFS
- NFS V4 (without delegation for shared file systems)
- NFS V3
- StorageTek ACSLS
- Solaris Cluster HA-QFS
- Solaris Cluster HA-SAM (SAM failover). It is only supported in a two host active/passive configuration. There are several configuration restrictions regarding this feature and the SAM Administration Guide should be read prior to using this feature.
- Solaris Cluster HA-NFS with HA-SAM. NFS v3 and v4 were tested. NFS file systems must be exported from the current Oracle HSM MDS. Minimum code levels for using HA-NFS with HA-SAM are Solaris 11.1 SRU 19.6 and SAM-QFS 5.3.
- Solaris Cluster HA-SAMBA with HA-SAM. File systems exported via SAMBA must be exported from the current MDS. Minimum tested code levels: Oracle HSM 6.1, OSC 4.3, Solaris 11.3.
- Solaris Cluster HA-Oracle
- Solaris Cluster Scalable (Shared) QFS with Oracle Real Application Cluster (Oracle RAC)
- Oracle Storage Cloud Object and Archive Service is supported as a new tier of storage starting with the 6.1-01 patch. See the README.txt for additional details.
- Solaris mixed architecture (SPARC with x64) Shared QFS configurations requires use of EFI disk labels.
- Shared QFS should be implemented using a private or dedicated ethernet switched network for metadata transmission. Performance problems can occur if this network is not private.
- For Shared QFS configurations of 10 hosts (nodes) or greater it is recommended that applications not be run on the metadata server to provide more predictable client performance.
- Sun StorageTek Enterprise Backup to a Oracle HSM file system.
- Veritas NetBackup to a SAM-QFS file system.
- Clients outside of Solaris Cluster in Solaris Cluster configurations. This includes clients of mixed architectures (SPARC with x64). This capability is only supported with the ma file system type.
- Zones (containers). Only one non-global zone per file system is supported. There are several configuration restrictions regarding this feature and the Oracle HSM Installation and Configuration Guide should be read prior to using this feature. For Solaris Cluster, Oracle RAC is supported in a Zone Cluster.
- Oracle VM Server for SPARC (previously known as Sun Logical Domains, or LDOMs) is supported with Oracle HSM, with the following restrictions:
- Minimum of 4 cores assigned to the domain.
- Suggested minimum of 24GB of RAM.
- SAM-QFS MDS must boot from a physical device, and therefore it must have at least 1 PCI root complex.
- Oracle VM Server virtual disks (vdisks) are supported for the QFS file system (server and client). Other non-Oracle disk I/O virtualization is not supported for LUNs which are used in a QFS file system.
- Network virtualization is supported.
- Tape devices must be attached via non-virtualized PCI slots attached to the Oracle HSM servers or clients.
- QFS clients may boot from a virtualized disk.
- Oracle VM for x86 is not supported.
- For Solaris Cluster, virtual storage devices used by Shared QFS must be backed by whole SCSI FC LUN disk arrays, and must not be shared with any other guest domain on the same server; virtualized partial LUN disk arrays are not supported.
- If running SAM with Shared QFS, SAM must be run on the metadata server.
- A version 1 file system cannot be upgraded to a version 2 file system.
- SAM-Remote servers and clients must be running the same revision level of Oracle HSM.
- Remote disk archiving servers and clients must be running the same revision level of Oracle HSM.
- Shared QFS clients must be running within one software revision of the Shared QFS metadata server. So if the Shared QFS metadata server is at revision level "N", the Shared QFS clients must be at revision level N or N-1. (Note that a Shared QFS client will be at revision level N+1 only when performing the first step of a rolling upgrade where the potential metadata server is upgraded.)
- Multi-reader servers and clients must be running the same revision of Oracle HSM or QFS software.
- Mixed architecture (SPARC with x64) metadata servers are not supported for failover purposes.
- Mixed architecture (SPARC with x64) multi-reader configurations are not supported.
- Online shrink. Online shrink is only supported for V2A ma file systems. Online shrink now includes support for Solaris Cluster.
- LTFS import feature requires Solaris 11.1 minimum.
- Extended attributes feature if desired on Linux clients, requires 2.6.32 Linux kernel minimum.
- WORM flags cannot be set from CIFS clients.
- The distributed I/O feature is limited to 9 clients.
- Multiple fibre channel paths to the T10000D tape drive is not supported with the "extended copy" or "StorageTek Direct mode" of media migration.
- The distributed I/O feature is not available on Solaris 11.4 (with patch 6.1-04).
- The administration GUI (also known as SAM-QFS manager) is removed as of patch 6.1-04. It was based on a Sun framework "Lockhart" which could not be kept up to date with security patches.
The following products, configurations, and features are not currently supported with Oracle HSM and QFS.
- Sun StorageTek Enterprise Backup of a SAM file system
- Veritas NetBackup of a SAM file system
- Instant Image (II)
- Sun StorageTek Network Data Replicator (SNDR)
- NFS V4 delegations with Shared QFS
- Solaris regular zones with Solaris Cluster
- Solaris Cluster HA-SAM with clients outside of Solaris Cluster
- Solaris Cluster Geographic Edition (SCGE)
- Segmented files on a shared file system
- Memory mapped segmented files
- Mixed architecture (SPARC with x64) metadata server failover
- Mixed architecture (SPARC with x64) multi-reader
- Oracle Linux 5 has reached end of life and will not be supported in future versions of Oracle HSM
- The Linux Shared QFS client doesn't support:
- access control lists (ACLs)
- being used as a NFS or Samba server
- sam-aio driver
- SAM-QFS Manager (GUI)
- 32 bit kernels on x64 systems
- forced unmount (umount -f)
- mdadm path failover
- Distributed I/O is not supported on Solaris 11.4 with patch 6.1-04.
- The administration GUI (SAM-QFS manager) is no longer supported with 6.1-04.
The following products and configurations were certified (tested) for Oracle HSM and QFS:
- Oracle Solaris 11 (base 11.0 release)
- Oracle Solaris 11.1
- Oracle Solaris 11.2
- Oracle Solaris 11.3
- Oracle Solaris 11.4 (with 6.1-04 patch)
- Note that Solaris 10 is no longer supported
The following versions of Linux were certified for Sun x64 64 bit architecture systems:
- Oracle Linux 5.9 - 5.11 (with default kernel, not unbreakable UEK kernel)
- RedHat 5.9 - 5.11 (via OL)
- Oracle Linux 6.3 - 6.9 (with default kernel, not unbreakable UEK kernel) (Patch 6.1-03 or higher is required to support OL versions higher than 6.5)
- RedHat 6.3 - 6.9 (via OL) (Patch 6.1-03 or higher is required to support RedHat versions higher than 6.5)
- Oracle Linux 7.0 - 7.3 (with default kernel, not unbreakable UEK kernel) (Patch 6.1-03 or higher is required to support OL versions higher than 6.5)
- RedHat 7.0 - 7.3 (via OL) (Patch 6.1-03 or higher is required to support RedHat versions higher than 6.5)
- SUSE 11 Service Pack 1 SMP SLES
4.2 Linux Shared QFS Clients (6.1-04 and beyond)
The following versions of Linux were certified for Sun x64 64 bit architecture systems:
- Oracle Linux 5 is no longer supported
- Oracle Linux 6.10 (with default kernel, not unbreakable UEK kernel)
- Oracle Linux 7.5 (with default kernel, not unbreakable UEK kernel). Oracle Linux beyond 7.5 are not tested.
- SUSE 11 Linux is no longer supported
- mdadm path failover is not supported. This is due to its use of a superblock at the end of each disk slice that Solaris is unaware of.
The following versions of HA-Oracle were certified with QFS and Solaris Cluster:
The following versions of Oracle RAC were certified with QFS and Solaris Cluster:
- Oracle RAC 22.214.171.124 for both SPARC and x86.
The following versions of OSC were certified with Shared QFS:
- OSC 4.3 with Solaris 11.3 with Oracle HSM 6.1-04. There an outstanding bug (30221971) that may prevent mounting / resource going online).
- OSC 4.4 with Solaris 11.4 with Oracle HSM 6.1-04 (recently added 26 Nov 2019). There are two outstanding bugs (30333778 and 30221971) that may prevent mounting / resource going online).
- If either of the above bugs are experienced, contact service with an SR for a special build that fixes the issue. These will be addressed in 6.1-05 patch.
- Clusters configured with Solaris Cluster should all be at the same major level of Solaris.
The following products were not certified or tested with Oracle HSM and QFS and thus the potential exists that a customer may have problems with them:
- Sun StorageTek Enterprise Backup of a QFS file system
- Veritas NetBackup of a QFS file system
- Veritas Cluster Services (VCS)
- Veritas Cluster Volume Manager (VCVM)
- Sun StorageTek Resource Management Suite
- Sun StorADE
- Any other Solaris Cluster agent with shared QFS configuration that hasn't been previously listed as supported
- HA-NFS with HA-SAM with code levels less than Solaris 11.1 SRU 19.6 or SAM-QFS 5.3.
Note that backup of QFS file system data should work. Backup products may not be aware of all of the QFS metadata extended attribute information, however, and this should be taken into account when backing up a QFS file system.