IBM FlashSystem V9000

Transcription

1 IBM FlashSystem V9000 Firmware Version Release Notes December 22, 2015 Contents 1.0 Applicable systems Product resources Bug severity legend Latest changes Issues fixed Known issues Currently supported specifications Upgrading firmware Overview of release Supported upgrade paths Preparing to upgrade Performing the upgrade Troubleshooting Stalled upgrade Failures during upgrade Contact information Release history... 7 Release Release Release Release Release Copyright notice Revision history... 11

2 1.0 Applicable systems These release notes apply to the following systems: IBM FlashSystem V9000, machine type and model (MTM) 9846-AC2 and 9846-AE2 IBM FlashSystem V9000, MTM 9848-AC2 and 9848-AE2 2.0 Product resources IBM FlashSystem V9000 product information resources guide users through the various features and components of the storage system, including usage and troubleshooting guides. To read about this storage system and learn how to use or troubleshoot, see the IBM Knowledge Center for IBM FlashSystem V9000 at or visit the IBM Redbooks website at for the IBM FlashSystem V9000 Product Guide. 3.0 Bug severity legend The following explains the bug severity ranking used in Section 4.1 for fixes made in the current release as well as fixes from previous releases listed in section 7.0: S1: Highest Recommend upgrade for all users as soon as possible. HIPER (highly pervasive) SAN Volume Controller (SVC) issues are also included. S2: Medium Recommend upgrade for all users at the next scheduled maintenance window. High importance SVC issues are also included. S3: Average Recommend upgrade at the next scheduled maintenance window for users experiencing these issues. All other users may wish to upgrade at the next scheduled maintenance window. Suggested SVC issues are also included. S4: Low Upgrade at the next scheduled maintenance window. May be performed at the discretion of the user if the issue is having a negative impact. S5: Lowest Upgrade is not necessary. This would include a mostly cosmetic or minor annoyance fix. 4.0 Latest changes After initial configuration of the hardware is complete, IBM strongly recommends that you make sure that your IBM FlashSystem firmware is up-to-date. Visit IBM Fix Central using the link below to see if any updates are available for your system Issues fixed The following are fixes that are included in this release and the severity level for each fix. Use these severity levels as described in Section 3.0 to aid in your decision to upgrade. Issues resolved in SVC related to the FlashSystem V9000 have also been included. Authorized Program Analysis Report (APAR) numbers are included for these issues. For more information on these issues, visit Fix Central for the SVC 7.6 release notes. S2 APAR-HU Batteries going on and offline can take node offline. S3 FLASH Potential I/O stall for more than 30 seconds during upgrade. APAR-HU Node assert observed when 16 Gb FC or 10 Gb FCoE adapter detects heavy network congestion. 2 Copyright IBM Corporation 2015, 2016

3 S4 FLASH Pools and I/O groups do not appear in the drop down when creating a HyperSwap volume if a long site name is configured. FLASH CLI help files were not translated in the release. 4.2 Known issues The following is a list of known issues and information related those these issues which exist for this release: The commands addvolumecopy and rmvolumecopy are not supported in the current release. Consequently, it is not possible to convert an existing volume to or from a HyperSwap volume by using these CLI commands or the management GUI. Administration of a HyperSwap volume can still be performed using the same CLI commands that were supported in version 7.5, configuring the component VDisks and active-active relationship that form the HyperSwap volume. Additionally, it is not currently supported to use the '-consistgrp' parameter on the new mkvolume command when creating a HyperSwap volume. A HyperSwap volume can still be assigned to a consistency group after it has been created using the management GUI or CLI. Attempting to use a feature which is not supported will return the following error: CMMVC7205E The command failed because it is not supported. This is a temporary restriction which will be lifted in a future release. Systems with more than 512 WWPNs in a single host object must reconfigure the host before upgrading to version Documentation is available for this issue at the following URL: Software update to version x may fail if the system date is set to any date prior to December 2014 with the following message: CMMVC6347E The specific update package cannot be installed on this hardware level Systems using Internet Explorer 11 may receive an erroneous "The software version is not supported" message when viewing the "Update System" panel in the GUI. Internet Explorer 10 and Firefox do not experience this issue. If using IP replication, please review the set of restrictions published in the Configuration Limits and Restrictions document for your product. Windows 2008 host paths may become unavailable following a Storwize V7000 node canister or SVC node replacement procedure. Documentation is available for this issue at the following URL: Intra-System Global Mirror not supported. More information is available at the following URL: Host Disconnects Using VMware vsphere Update 2 and vsphere 6.0. More information is available at the following URL: Version

4 4.3 Currently supported specifications SCSI-SAM-3 SCSI-SPC-3 SCSI-SBC-2 SCSI-FCP-3 SCSI-SRP FC-PH-3 FC-AL-2 SCSI Architecture Model v3 SCSI Primary Commands v3 SCSI Block Commands v2 Fibre Channel Protocol v3 SCSI RDMA Protocol v1 Fibre Channel Physical and Signaling Interface v3 Fibre Channel Arbitrated Loop v2 5.0 Upgrading firmware Use the following sections to perform code upgrades for your systems to the current release. Warning: Please read all the instructions below before upgrading. 5.1 Overview of release If you are upgrading to this release and your system is healthy, you can perform a Concurrent Code Upgrade (CCU). A CCU is a non-disruptive upgrade and is the preferred upgrade method. For general instructions on performing upgrades, refer to the FlashSystem Knowledge Center using the following URL: From this link, you can access the appropriate FlashSystem page through the following path: System Storage Flash Storage Flash high availability systems IBM FlashSystem V9000 Upgrading 5.2 Supported upgrade paths The following upgrade paths are supported for this release: Preparing to upgrade A CCU is a non-disruptive upgrade, which means that the system remains online throughout the process and that you can continue to access data normally. As a precaution, it is recommended that the upgrade occur during a time of reduced traffic. During the upgrade, the interface adapters in each control node are taken offline temporarily in order to be upgraded. This might impact performance or throughput. The impact is more noticeable under heavy load conditions. With a properly configured multi-path configuration, access to your data is maintained at all times. In order to ensure a successful, non-disruptive upgrade, you should verify that your interface ports are all online and all the system hardware is functioning normally. Ideally, you should have the following: All host interfaces should be online. An active multi-path configuration is required to ensure no loss of access during the upgrade. Information on this topic is available in IBM Knowledge Center under the Fibre Channel host attachments topic found using the following URL: nnel_cover.html 4 Copyright IBM Corporation 2015, 2016

5 All batteries should be online and charged at least 85%. Use the CLI command lsenclosurebattery to verify battery charge. If using encryption, ensure that both USB keys are inserted during the CCU. From the CLI, issue lsportusb to view if USBs have been inserted as an alternative to a physical check. From the GUI, navigate to Settings Security Encryption to view ports and encryption status. All hardware should be online and functioning normally. There should be no unfixed alerts in the event log (see the exceptions below). Important: Before you begin the upgrade, we recommend that you perform a backup of your data and a backup of the FlashSystem configuration. To back up the configuration, log into the cluster management IP address and issue the following command using an account with Administrator privileges: svcconfig backup Optionally, you can copy the configuration backup file from the FlashSystem to your workstation using secure copy (scp) on Linux or PuTTY secure copy (pscp.exe) on Windows as in the following examples: (Using Linux) scp username@cluster_ip:/dumps/svc.config.backup.*. (Using Windows) pscp username@cluster_ip:/dumps/svc.config.backup.*. Note: Do not ignore the periods shown above at the end of each command. In addition, replacement of italicized descriptions with appropriate information is required. 5.4 Performing the upgrade It is highly recommended that the upgrade be performed using the web-based cluster management interface known as the management GUI. Access this option through SettingsSystemUpdate System in the GUI. Using the management GUI, you will be prompted to install and run the software upgrade test utility and the software update package from a single GUI panel. The software upgrade test utility is designed to detect and warn of various conditions that might prevent a successful upgrade. The upgrade test utility can be run either during the upgrade using the Update and Test button or by itself with the Test only button. Once the utility finishes running certain checks for vulnerabilities, the software update utility will allow you to upload the firmware file and start the upgrade. For more detailed instructions and information for using the upgrade utility from the CLI or the GUI, see the release notes for the upgrade utility, which is located on Fix Central. The upgrade can proceed successfully without the Call Home feature enabled, but IBM highly recommends that the feature be enabled. The upgrade can also be performed using the applysoftware command in the CLI. This requires that you manually upload the latest release to the/upgrade directory on the cluster management node. The upgrade test utility must be run before the upgrade can start. Version

6 5.5 Troubleshooting Use the following sections to troubleshoot problems that may occur during the upgrade process Stalled upgrade If the upgrade completion percentage does not increase for more than two hours, the upgrade may have stalled. Upgrade status is viewed by issuing the lsupdate CLI command or by going to Settings System Update System in the GUI. Both show a Stalled status. If the upgrade stalls or if you have alerts which cannot be cleared, contact IBM Support. Support contact information is provided in Section 6.0 of this document Failures during upgrade You may get a battery or quorum alert during upgrade. These alerts are expected and should be automatically cleared when the upgrade is completed. They may be visible from the Events view of the management GUI if the filter is set to "Show All," but they should no longer appear in the Recommended Actions, Unfixed Messages, or Alerts views. If you see unfixed battery or quorum alerts after an upgrade is complete, contact IBM Support. If the upgrade has stalled due to a hardware failure, you ll see the current status of Hardware Failed. If you suspect a hardware failure, issue the lsupdate command from the management CLI to confirm the state of your system. This command shows that the system is in a hardware_failed state and the event log contains a "System upgrade suspended" event. You may resume the upgrade by issuing the command applysoftware -resume force for the following conditions: PSU unsupported events Battery fault type 1 events that are fixed and online according to the CLI command lsenclosurebattery Fan events If the upgrade cannot be resumed or you have other alerts which cannot be cleared, contact IBM Support. All releases include a battery reconditioning feature which calibrates the gauge that reports the amount of charge on the batteries. On systems that have been installed for 10 months or more or systems that have experienced several power outages, the recommendation to run "battery reconditioning" will appear in the event log shortly after upgrading. This is normal. Use the management GUI to run the DMP for this error. 6 Copyright IBM Corporation 2015, 2016

7 6.0 Contact information Call IBM at IBM-SERV ( ). To find contact information for a specific region, visit the IBM directory of worldwide contacts at Release history The following sections include a list of all fixes and improvements for previous releases. Reference Section 3.0 for explanations of each severity level. Release The following issues were fixed in release : FLASH HIPER (highly pervasive): Potential undetected data corruption may occur due to a low probability race condition. The race condition has been observed on a system with a specific workload that is doing 1 to 2 GB/s of read operations with 250 MB/s of write operations. The write operations were less than 4K in size. (S1) APAR-HU Stop with access resulted in a node assert after a recoveryvdiskbysystem command. (S1) APAR-HU Multiple node asserts in IO group after starting Remote Copy. (S1) APAR-HU Multiple node asserts when removing a Global Mirror relationship with secondary volume that has been offline. (S1) APAR-HU Large increase in response time of Global Mirror primary volumes due to intermittent connectivity issues. (S1) APAR-HU Multiple node asserts when space efficient volumes are running out of capacity. (S1) FLASH Repeated interface panics causes an incorrect failure. (S2) APAR-HU Increased CPU utilization on nodes from Easy Tier processing. (S2) APAR-HU Node assert due to inconsistent Easy Tier status when Easy Tier is disabled on all managed disk group. (S2) APAR-HU Both nodes in a single IO group of a multi IO group system can assert due to misallocation of volume stats entries. (S2) APAR-HU FlashCopy volume offline due to a cache flush issue. (S2) APAR-HU SVC FC driver asserts when it receives an unsupported but valid FC command. (S2) APAR-HU Battery can charge too quickly on reconditioning and take node offline. (S2) APAR-HU Single node assert when removing an Mdisk group that had EasyTier activated. (S2) APAR-HU Loss of access to data when removing volumes associated with a GMCV relationship. (S2) APAR-HU Performance impact on read pre-fetch workloads due to cache pre-fetch changes. (S2) APAR-HU FlashCopy volumes may go offline during an upgrade. (S2) APAR-HU CCU checker causes both nodes to assert. (S2) APAR-HU Gb HBA causes mutiple node asserts when unexpected FC frame content received. (S2) APAR-HU Increase in workload may cause nodes to assert. (S2) APAR-HU Unable to remove ghost mdisks. (S2) FLASH A node timeout causes the Flash to fail. (S3) FLASH FC interface timeout issue results in a temporary error. (S3) FLASH Unexpected node assert. (S3) FLASH Failed encryption validation causes a VM timeout. (S3) FLASH Flash failure due to Gateway to node CRC errors. (S3) FLASH Flash card failures are a result of an unexpected power off. (S3) FLASH Mitigate flashcard encryption error. (S3) FLASH Performance improvements made. (S3) Version

8 APAR-HU Single node assert on login attempt with incorrect password issued. (S3) APAR-HU When stopping a GMCV relationship clean up process at secondary site hangs to the point of a primary node assert. (S3) APAR-HU SVC reports invalid counters causing TPC to skip sample periods. (S3) APAR-HU Hang up of GM IO stream can effect MM IO in another remote copy stream. (S3) APAR-HU FlashCopy performance degraded. (S3) APAR-HU Emulex firmware paused causes single node assert. (S3) APAR-HU Single node assert when receiving frame errors on 16GB Fibre Channel adapters. (S3) APAR-HU Single node assert when concurrently creating new volume host mappings. (S3) APAR-HU Support for Fujitsu Eternus DX100 S3 controller. (S3) APAR-HU SNMP and Call home stops working when node reboots. (S3) APAR-HU Volume hosted on Hitachi controllers show high latency due to high IO concurrency. (S3) APAR-HU Single node asserts due to FlashCopy chaining. (S3) APAR-HU Single node assert during code upgrade. (S3) APAR-IC The exact file requested is missing when copying statistic files between nodes presents event ID (S3) APAR-IT Freeze time update delayed after reduction of cycle period. (S3) FLASH The upgrade utility does not report a failed drive. (S4) FLASH Service manager panic experienced during upgrade. (S4) FLASH Properly cabled fixed configurations always raise a false "Minimum cabling rules not met alert. (S4) FLASH Stalled upgrade reports an upgrade failure error even after the upgrade completes successfully. (S4) FLASH Incomplete chenclosurecanister command can cause the nodes to assert. (S4) FLASH A node assert is experienced during upgrade. (S4) FLASH A mandatory parameter is not included in the ping CLI help. (S5) Release The following issues were fixed in release : FLASH Include all of the following relevant fixes from the SVC release, version : (S1) HU Loss of access to data when a volume that is part of a Global Mirror Change Volumes relationship is removed with the 'force' flag. HU Both nodes shutdown when power is lost to one node for more than 15 seconds. HU Multiple node asserts leading to loss of access to data when the link used for IP Replication experiences packet loss and the data transfer rate occasionally drops to zero. HU Single node assert due to HBA firmware behavior. HU The serial number value displayed in the GUI node properties dialog is incorrect. FLASH x1 fixed configurations raise a false Minimum cabling rules not met event. (S2) FLASH Running the canister intercommunication error DMP (error code 1007) yields an Unable to Proceed message. (S2) FLASH Possible upgrade issues can take place for enclosures with encrypted arrays. (S2) FLASH Remediate vulnerabilities in Java (CVE , CVE , CVE , and CVE ). (S2) FLASH When trying to install some packages, a Error in verifying the signature of the update package message is produced. (S3) FLASH If a CCL is performed while the flash modules are in a Waiting state, the modules can write over used flash and this therefore causes a fault when read. (S4) 8 Copyright IBM Corporation 2015, 2016

9 Release The following issues were fixed in release : FLASH-10119, HIPER (highly pervasive): Potential undetected data corruption may occur from an interface error. In rare cases, the FlashSystem V9000 can write inconsistent data to a host. This is considered a highly pervasive problem involving firmware versions 7.4.x.x. (S1) FLASH Remediate security vulnerabilities in OpenSSL (CVE , CVE , CVE , and CVE ). (S1) FLASH Remediate security vulnerabilities in Java CPU (CVE , CVE , and CVE ). (S1) FLASH Remediate security vulnerability in Apache Struts (CVE ). (S1) FLASH-12692, FLASH Improve canister replacement DMPs. (S2) FLASH Unable to add storage enclosure at a higher version than the current configuration. (S2) FLASH Both canisters incorrectly downgrade at the same time during a failed CCU (PMR). (S2) FLASH-10600, FLASH Adding a second building block with down-level firmware to existing cluster does not complete (PMR ). (S2) FLASH MDisk reports degraded ports, but the DMP does not offer a real fix to the issue (PMR). (S2) FLASH Canister to canister communication, error 1007 identifies the incorrect canister for replacement (PMR). (S2) FLASH Node error , Inter-canister PCIe link failure is seen too often. (S2) FLASH Scaled up configuration encryption validation only runs on one enclosure. (S2) FLASH False disk encryption error (085158) reported (PMR). (S2) FLASH Errors incorrectly blame flash module instead of quorum device. (S2) FLASH Unexpected Input/output (I/O) errors causes controller node assertions every 30 minutes (PMR). (S2) FLASH Enclosure canister failed to come online after a cold start (PMR). (S2) FLASH Candidate drives are exposed when they are not ready for I/O resulting in false failures during FRU replacements. (S2) FLASH The RAID controller incorrectly marked as failed instead of flashcard in a particular double flashcard failure scenario. (S2) FLASH Unexpected canister power off can cause a warm start on remaining canister. (S2) FLASH Fan fault events incorrectly appear during the enclosure boot upgrade. (S2) FLASH I/O latency spikes due to periodic encryption validation. (S3) FLASH Move key validation from 30 to 1440 minutes. (S3) FLASH-9731, FLASH Flash sustained write performance improvements. (S3) FLASH Flash endurance improvements. (S3) FLASH There is no GUI option to add new expansion enclosure to existing pool when scaling up. (S3) FLASH After a port between AC2 and AE2 goes down and is recovered, it incorrectly appears as a phantom controller (PMR). (S3) FLASH Issuing cpenclosuredumps following cleardumps does not copy any file. (S3) FLASH Unsupported 4th adapter in the AC nodes results in service 564 state (PMR). (S3) FLASH Issuing lsdependentvdisks -drive 0 causes recoverable a warm start on node (PMR). (S3) FLASH There is an issue on performance reporting with 16Gb FC. (S3) FLASH Stale interface logins are not removed on failure. (S3) FLASH-11128, FLASH-12620, FLASH-12690, FLASH Canister replacement DMP improvements. (S3) FLASH Canister warm start DMP improvement. (S3) Version

10 FLASH Flashcard reporting incorrect temperature on node timeout resulting in false critical temperature failure. (S4) FLASH Flash improvements in error path identification. (S4) FLASH Logging improvements. (S4) FLASH GUI incorrectly shows member drives with Suitability "Wrong Slot" (PMR). (S4) FLASH Issuing clearenclosuredumps works only for one canister with new -canister option. (S4) FLASH Cabling code not working as designed. (S4) FLASH Management controller packet handling improvement. (S4) FLASH Build string truncated in lsservicestatus and the Service Assistant GUI. (S4) FLASH Improve canister node error reporting. (S4) FLASH Issuing lsdumps prefix with an invalid directory causes a node failover. (S4) FLASH Improve logging. (S5) FLASH Snap output is missing results from lsnodebattery. (S5) FLASH svc_snap performance improvements. (S5) Release The following issue was fixed in release : FLASH This release includes one fix which addresses an array rebuild omission issue found in the IBM FlashSystem V9000 for 5.7 TB flash modules. (S1) Release The following issues were fixed in release : FLASH Improved handling of retention scan. (S1) FLASH Fixed canister warm start connection issue. (S2) FLASH Node goes into service state with error 564 after exiting service state with error 657 due to double battery removal. (S2) FLASH Improved handling of canister failover during a CCU for canisters at different firmware levels. (S2) FLASH Improved array reaction to simultaneous flash and interface failure. (S2) FLASH Encryption validation supported for larger environments. (S3) FLASH Improvements made to interface behavior during cable pulls and host connects and disconnects. (S3) FLASH Issuingcharraymember command can cause node failover. (S3) FLASH Improved behavior of batteries upon re-insertion. (S3) FLASH Changed default snap to not include historical stats to their reduce size. (S4) FLASH Call home event spamming on successful read retries fixed. (S4) FLASH Interface DRAM ECC failure incorrectly flags flash failure fixed. (S4) FLASH-10079, Improved output displayed for lsarraymemberprogress command. (S4) FLASH Improved Flash during CCU from code version (S4) FLASH Fixed array creation in the management GUI s setup wizard. (S4) FLASH Adjusted enclosure WWPN displayed information on the DMP for error code (S4) FLASH Improved error reporting for enclosure canister failure. (S4) FLASH Improved CLI reporting after canister failover. (S4) FLASH Irrelevant cabling events are no longer reported after upgrading. (S4) FLASH Improved flash testing on reboots. (S5) FLASH Panel name now appears on initial navigation to Settings Security or Settings System. (S5) FLASH Improved interface BW statistics reporting. (S5) FLASH Improvedapplysoftware with unmanaged enclosure. (S5) 10 Copyright IBM Corporation 2015, 2016

11 FLASH FC port view improved on the GUI. (S5) FLASH Fixed reporting after deleting an MDisk. (S5) FLASH Upgrade system view is now more reliable. (S5) 8.0 Copyright notice IBM, the IBM logo, and ibm.com are trademarks or registered trademarks of IBM Corporation in the United States, other countries, or both. These and other IBM trademarked terms are marked on their first occurrence in this information with the appropriate symbol ( or ), indicating US registered or common law trademarks owned by IBM at the time this information was published. Such trademarks may also be registered or common law trademarks in other countries. A current list of IBM trademarks is available on the Web at The following terms are trademarks of other companies: Linux is a registered trademark of Linus Torvalds in the United States, other countries, or both. Microsoft, Windows, Windows NT, and the Windows logo are trademarks of Microsoft Corporation in the United States, other countries, or both. Java and all Java-based trademarks and logos are trademarks or registered trademarks of Oracle and/or its affiliates. 9.0 Revision history The following table outlines the revision history of this document. Document version Date Revision details 1.1 January 6, 2016 Section 4.2: Added issue APAR-HU Section 7.0: Removed issue APAR-HU Version