Release Note for systems built with IBM Spectrum Virtualize


This is the release note for the 8.5.3 release and details the issues resolved between 8.5.3.0 and 8.5.3.1. This document will be updated with additional information whenever a PTF is released.

Note: This release is a Non-Long Term Support (Non-LTS) release. Non-LTS code levels will not receive regular fix packs. Fixes for issues introduced in a Non-LTS release will be delivered in a subsequent Non-LTS or LTS release. If issues are encountered the only resolution is likely to be upgrade to a later LTS or Non-LTS release.

This document was last updated on 28 October 2024.

  1. New Features
  2. Known Issues and Restrictions
  3. Issues Resolved
    1. APARs Resolved
  4. Useful Links

Note. Detailed build version numbers are included in the Update Matrices in the Useful Links section.


1. New Features

The following new features have been introduced in the 8.5.3 release:

2. Known Issues and Restrictions

Note: For clarity, the terms "node" and "canister" are used interchangeably.
Details Introduced

Customers planning to upgrade to v8.4.0 or later should be aware that an update to OpenSSH has terminated support for all DSA keys. An update to OpenSSL has also terminated support for 1024-bit RSA keys used in SSL certificates.

Customers currently using DSA public keys for SSH access will need to generate new keys using alternative ciphers, such as RSA or ECDSA. If using RSA public keys for SSH access, it is recommended to use keys of 2048 bits or longer.

Customers currently using 1024-bit RSA keys in SSL certificates will need to generate new SSL certificates using 2048-bit RSA keys or ECDSA keys. This applies not only to the system certificate, but also to any SSL certificates used by external services such as LDAP servers.

8.4.0.0

There is an existing limit on the number of files that can be returned by the CLI of approximately 780 entries. In many configurations this limit is of no concern. However, due to a problem with hot-spare node I/O stats files, 8-node clusters with many hardware upgrades or multiple spare nodes may see up to 900 I/O stats files. As a consequence the data collector for Storage Insights and Spectrum Control cannot list or download the required set of performance statistics data. The result is that there are many gaps in the performance data, leading to errors with the performance monitoring tools and a lack of performance history.

The workaround is to remove the files associated with spare nodes or previously/updated hardware using the cleardumps command (or to cleardumps the entire iostats directory).

This is a known issue that will be lifted in a future PTF. The fix can be tracked using APAR HU02403.

8.4.0.0

Customers using Spectrum Control v5.4.3 or earlier may notice that IP port status is incorrectly shown as "Unconfigured".

This issue will be resolved by a future release of Spectrum Control.

8.4.2.0

Upgrade to 8.5.1 and higher is not currently supported on systems with compressed volume copies in Data Reduction Pools, to avoid SVAPAR-105430.

On a small proportion of systems, this can cause a node warmstart when specific data patterns are written to compressed volumes. This restriction will be lifted in a future release.

8.5.1.0

If a Secured IP Partnership fails to connect properly, the system will log a 2020 error. However the Fix Procedure for the 2020 error is unable to guide the user to fix the problem.

If you are experiencing a 2020 error and can't diagnose the problem directly, please contact IBM support.

8.5.2.0

Systems that upgrade from pre-8.5.2 to 8.5.2 or later will now default to 5 minutes for the performance statistics file generation, regardless of any previous settings.

Note that only the first upgrade to a code level containing this change is affected.

8.5.2.0

Customers using IP Replication should be aware that the improved performance may produce more data over replication/WAN link.

Customers should ensure that the bandwidth of their replication/WAN link has sufficient capacity to avoid the possibility of network congestion.

8.5.2.0

Upgrade from 8.5.2.1 is not supported if replication policies exist, due to APAR HU02546.

Remove all replication policies before upgrading from 8.5.2.1. If this is not possible, contact your support representative.

8.5.2.1

Systems with iSCSI two-way CHAP authentication enabled cannot upgrade to 8.5.3.

To enable the upgrade to proceed, disable two-way CHAP authentication, complete the upgrade, and then re-enable two-way CHAP authentication.

8.5.3.0

Systems, with NPIV enabled, presenting storage to SUSE Linux Enterprise Server (SLES) or Red Hat Enterprise Linux (RHEL) hosts running the ibmvfc driver on IBM Power can experience path loss or read-only file system events.

This is cause by issues within the ibmvfc driver and VIOS code.

Refer to this troubleshooting page for more information.

n/a
If an update stalls or fails then contact IBM Support for further assistance n/a
The following restrictions were valid but have now been lifted

Customers using Storage Insights with systems running policy-based replication will see alerts for recovery volumes which are intentionally offline with the policy-based replication design.

This issue has now been resolved under Storage Insights APAR IT42347.

8.5.2.1

Applications that use 'lsportip' such as Veeam to retrieve iSCSI target IP address information from a Spectrum Virtualize system will fail.

The 'lsportip' command was deprecated in Spectrum Virtualize code release 8.4.2, so the host server is unable to detect that the Spectrum Virtualize system is iSCSI capable, resulting in failures

In the case of Veeam, there is no support for storage snapshot integration with IBM FlashSystem running Spectrum Virtualize 8.4.1 or newer, however they plan to add support for newer releases of Spectrum Virtualize in the next major release.

This issue has now been resolved as Veeam Backup and Replication Version 12 no longer has this restriction

8.4.1.x

3. Issues Resolved

This release contains all of the fixes included in the 8.5.2.2 release, plus the following additional fixes.

A release may contain fixes for security issues, fixes for APARs or both. Consult tables below to understand the complete set of fixes included in the release.

3.1 Security Issues Resolved

Security issues are documented using a reference number provided by "Common Vulnerabilities and Exposures" (CVE).
CVE Identifier Link for additional Information Resolved in

3.1 APARs Resolved

Show details for all APARs
APAR Affected Products Severity Description Resolved in Feature Tags
SVAPAR-82950 FS9500, SVC High Importance If a FlashSystem 9500 or SV3 node had a USB Flash Drive present at boot, upgrading to either 8.5.0.7 or 8.5.3.0 may cause the node to become unresponsive. Systems already running 8.5.0.7 or 8.5.3.0 are not affected by this issue (show details)
Symptom Loss of Redundancy
Environment FlashSystem 9500 or SV3 node with a USB Flash Drive present
Trigger Upgrade to 8.5.3.0
Workaround None
8.5.3.1 Reliability Availability Serviceability
HU02441 & HU02486 All Critical Safeguarded Copy with DRP can cause node warmstarts and mdisk timeouts (show details)
Symptom Loss of Access to Data
Environment Systems running v8.4.2 or later using Safeguarded Copy with DRP
Trigger None
Workaround None
8.5.3.0 Data Reduction Pools, Safeguarded Copy & Safeguarded Snapshots
HU02541 All Critical In some circumstances, the deduplication replay process on a data reduction pool can become stuck. During this process, IO to the pool is quiesced and must wait for the replay to complete. Because it does not complete, IO to the entire storage pool hangs, which can eventually lead to a loss of access to data. (show details)
Symptom Single Node Warmstart
Environment Systems running DRP - 8.4.2, 8.5+
Trigger None
Workaround Warmstart the nodes
8.5.3.0 Data Reduction Pools, Deduplication
HU02551 All Critical When creating multiple volumes with a high mirroring sync rate, an node warmstart maybe triggered due to internal resource constraints (show details)
Symptom Single Node Warmstart
Environment Systems running 8.5.0, 8.5.1, 8.5.2
Trigger Creating muliple volumes with high mirroring sync rate
Workaround Lower the sync rate to 100 when creating multiple volumes
8.5.3.0 Global Mirror, Global Mirror With Change Volumes, Metro Mirror
HU02561 All Critical If there are a high number of FC mappings sharing the same target, the internal array that is used to track the FC mapping is mishandled, thereby causing it to overrun. This will cause a cluster wide warmstart to occur (show details)
Symptom Multiple Node Warmstarts
Environment System running v8.3.1,v8.4.0,v8.4.1,v8.4.2,v8.5.0
Trigger Cascaded flashcopy mappings and one of the flashcopy target volumes is the source of 255 flashcopy mappings
Workaround None
8.5.3.0 FlashCopy
HU02563 All Critical Improve dimm slot identification for memory errors (show details)
Symptom Single Node Warmstart
Environment System running v8.4.0,v8.4.1,v8.4.2,v8.5.0,v8.5.1,v8.5.2
Trigger Bad memory module
Workaround Work with IBM Support to replace the DIMM that reported the uncorrectable error and replace it
8.5.3.0 Reliability Availability Serviceability
HU02567 All Critical Due to a low probability timing window, FlashCopy reads can occur indefinitely to an offline Vdisk. This can cause host write delays to flashcopy target volumes that can exceed 6 minutes (show details)
Symptom Loss of Redundancy
Environment All
Trigger Offline vdisk in the flashcopy dependency chain
Workaround None
8.5.3.0 FlashCopy
HU02525 FS5000, FS5100, FS5200, FS7200, FS7300, FS9100, FS9200, FS9500, SVC High Importance Code versions 8.4.2.x, 8.5.0.0 - 8.5.0.5 and 8.5.1.0 permitted the use of an iSCSI prefix of 0. However, during an upgrade to 8.5.x, this can prevent all iSCSI hosts from re-establishing iSCSI sessions, thereby causing access loss (show details)
Symptom None
Environment Systems running v8.4.2,v8.5.0,v8.5.1
Trigger iSCSI prefix of 0
Workaround Change all hosts with a prefix of 0 before upgrading
8.5.3.0 Hosts, iSCSI
HU02534 All High Importance When upgrading from 7.8.1.5 to 8.5.0.4, PowerHA stops working due to SSH configuration changes (show details)
Symptom None
Environment Systems running 8.5 and higher
Trigger None
Workaround The PowerHA script 'cl_verify_svcpprc_config' can be changed to use the actual username instead of 'admin'
8.5.3.0 Reliability Availability Serviceability
HU02549 All High Importance When upgrading from a lower level, to 8.5 or higher for the first time, an unexpected node warmstart may occur that can lead to a stalled upgrade (show details)
Symptom Single Node Warmstart
Environment Systems upgrading from a lower level to 8.5 or higher
Trigger First time upgrade to 8.5 or higher from a lower level
Workaround None
8.5.3.0 System Update
HU02555 All High Importance A node may warmstart if the system is configured for remote authorization, but no remote authorization service, such as LDAP, has been configured (show details)
Symptom Single Node Warmstart
Environment Systems running 8.5.x onwards
Trigger Attempted login by a user that is not a locally configured user such as LDAP
Workaround Log in to the node with a known local user (i.e. superuser) then run 'svctask chauthservice -enable no -type ldap' to correct the auth inconsistency
8.5.3.0 LDAP
HU02562 All High Importance A node can warmstart when a 32 Gb Fibre Channel adapter receives an unexpected asynchronous event via internal mailbox commands. This is a transient failure caused during DMA operations (show details)
Symptom Loss of Access to Data
Environment Any system with a 32 Gb Fibre Channel adapter installed and running code level 8.4.0.4 or higher
Trigger A 32 Gb Fibre Channel adapter receives an unexpected asynchronous event via internal mailbox commands
Workaround None
8.5.3.0
HU02569 All High Importance Due to a low-probability timing window, when processing I/O from both SCSI and NVMe hosts, a node may warmstart to clear the condition (show details)
Symptom Single Node Warmstart
Environment Systems running 8.5.x when processing I/O from both SCSI and NVMe hosts
Trigger Stale NVME command encountered in the control block
Workaround Upgrade to 8.5.3.0
8.5.3.0 Host Cluster, Hosts, NVMe, SCSI Unmap, iSCSI
HU02580 All High Importance If FlashCopy mappings are force stopped, and the targets are in a remote copy relationship, then a node may warmstart (show details)
Symptom Single Node Warmstart
Environment Systems running 8.5.2 with FlashCopy mappings that have been force stopped and the targets are in a remote copy relationship
Trigger FlashCopy mappings that have been force stopped and the targets are in a remote copy relationship
Workaround Delete and recreate the Flashcopy mappings, or stop the Flashcopy mappings without using the force option
8.5.3.0 FlashCopy
HU02581 All High Importance Due to a low probability timing window, a node warmstart might occur when I/O is sent to a partner node and before the partner node recognizes that the disk is online (show details)
Symptom Single Node Warmstart
Environment All systems
Trigger Failing I/O on the remote node causes the local I/O to fail, believing the path is about to go offline, but it never does, thereby causing I/O to be queued indefinitely, leading to a node warmstart.
Workaround None
8.5.3.0 Cache
HU02583 All High Importance FCM drive ports maybe excluded after a failed drive firmware download. Depending on the number of drives impacted, this may take the RAID array offline (show details)
Symptom Loss of Access to Data
Environment Systems running v8.3.1
Trigger Drive firmware upgrade
Workaround Physical reseat or power cycle the drive
8.5.3.0 Drives
HU02528 All Suggested When upgrading to 8.5.0 or higher, a situation may occur whereby a variable is not locked at the correct point, resulting in a mismatch. The system code detects this and initiates a warmstart to reset any erroneous values (show details)
Symptom Single Node Warmstart
Environment Systems that have upgraded to v8.5
Trigger Can only occur during upgrade to 8.5 or higher
Workaround None
8.5.3.0 Reliability Availability Serviceability
HU02553 FS9500, SVC Suggested Remote copy relationships may not correctly display the name of the vdisk on the remote cluster (show details)
Symptom None
Environment Any FS9500 or SV3 node running 8.4.2 or later, and using remote copy
Trigger None
Workaround None
8.5.3.0 Global Mirror, Global Mirror With Change Volumes, Metro Mirror
HU02559 All Suggested A GUI resource issue may cause an out-of-memory condition, leading to the CIMOM and GUI becoming unresponsive, or showing incomplete information (show details)
Symptom None
Environment Systems running v8.3.1,v8.4.0,v8.4.1,v8.4.2,v8.5.0
Trigger None
Workaround Restart the cimserver service or tomcat service
8.5.3.0 Graphical User Interface
HU02579 All Suggested The GUI 'Add External iSCSI Storage' wizard does not work with portsets. The ports are shown but are not selectable (show details)
Symptom None
Environment Any system running V8.4.2,V8.5.0,V8.5.1,V8.5.2, and using IP portsets
Trigger None
Workaround Use the command line to configure the external iSCSI connection
8.5.3.0 Graphical User Interface, iSCSI

4. Useful Links

Description Link
Product Documentation
Update Matrices, including detailed build version
Support Information pages providing links to the following information:
  • Interoperability information
  • Product documentation
  • Limitations and restrictions, including maximum configuration limits
Spectrum Virtualize Family of Products Inter-System Metro Mirror and Global Mirror Compatibility Cross Reference
Software Upgrade Test Utility
Software Upgrade Planning