Isilon: How to perform simultaneous drive firmware updates manually using Drive Support Package 1.19 and newer

NOTE: Please read the Known Limitations section for information on potential issues with this procedure before following the below steps! NOTE: Following this procedure will reboot the node! If you would prefer to update drive firmware in a non-disruptive (rolling) fashion, please follow the normal drive firmware upgrade instructions included in the DSP release notes. The steps provided below will allow you to update the drive firmware on all drives in a node simultaneously. They will work on both joined and unjoined (unconfigured) nodes; if working on an unconfigured node, press 'CTRL-\' (CTRL-backslash) to break out of the configuration wizard to the CLI prompt. If working on an unconfigured node that will be part of a compliance mode cluster, we recommend completing these steps before switching the node to compliance mode. This procedure is similar to, and replaces, the deprecated Drive Firmware Package (DFP) method of updating all drives simultaneously. Starting in DSP 1.19, this will be the only method for updating drive firmware on OneFS version 7.1.0 and older. This procedure also deprecates the 'No_SSD' branch of the Drive Support Package; if you are running OneFS versions 7.1.1.0 - 7.1.1.3 or 7.2.0.0 - 7.2.0.1, you must use this procedure to update drive firmware using DSP 1.19 or newer. Please see ETA 200097: Isilon OneFS 7.1.1.0 - 7.1.1.3 and 7.2.0.0 - 7.2.0.1: Attempts to upgrade SSD drive firmware using an Isilon Drive Support Package may result in data loss on clusters that have the L3 cache feature enabled for details on this issue. 1. First we need to verify that there is free disk space in the /var/crash partition. # df -h /var/crash We recommend having at least 100MB free space in /var/crash before proceeding. If /var/crash is full, and you can identify stale troubleshooting data (such as old core files), remove them to make space; if you need assistance in deciding what to clear, please contact Isilon Support for assistance. 2. Download the latest version of the Drive Support Package (DSP) from support.emc.com and place it on the node in the /var/crash folder. 3. If the node is in a cluster, and the cluster does not already have the latest DSP installed, make a copy of the downloaded DSP file and install it on the cluster using the steps below. If the node is in a cluster that already has the latest DSP installed, skip this step and proceed to step 4. If you are working on an unconfigured node, save a copy of the downloaded DSP and install it on the cluster once the node has been configured or joined, using the normal installation instructions in the DSP release notes. NOTE: If the cluster is running OneFS 7.1.1.0 - 7.1.1.3 or 7.2.0.0 - 7.2.0.1, the DSP will refuse to install to protect the node from the L3 cache issue described in ETA 200097: Isilon OneFS 7.1.1.0 - 7.1.1.3 and 7.2.0.0 - 7.2.0.1: Attempts to upgrade SSD drive firmware using an Isilon Drive Support Package may result in data loss on clusters that have the L3 cache feature enabled. Skip this step and proceed to step 4, and consider upgrading OneFS to a newer version ASAP to eliminate this issue. # cp /var/crash/Drive_Support_<version>.tgz /ifs/data # tar -zxf /ifs/data/Drive_Support_<version>.tgz # isi_dsp_install /ifs/data/Drive_Support_<version>.tar 4. Next, we need to extract the DSP contents to /var/crash/dsp: # cd /var/crash # tar -zxf Drive_Support_<version>.tgz # mkdir /var/crash/dsp # tar -C /var/crash/dsp -xf Drive_Support_<version>.tar 5. Now we can verify that there are drives in the node that need updates: # cd /var/crash/dsp # ./isi_update_drive_firmware 6. When ready, run the following command to perform a simultaneous drive firmware update: NOTE: This command will reboot the node! # ./isi_update_drive_firmware -up When the update is complete, the node will reboot. Once the node finishes booting, log back in and proceed to step 7. If you are working on an unconfigured node, it will reboot back into the configuration wizard; press 'CTRL-\' to go back to the CLI prompt. 7. Now we can verify that all drives in the node were updated: # cd /var/crash/dsp # ./isi_update_drive_firmware If all updates were installed successfully, the response should be: No updates available 8. If the update was successful, we can clean up the extracted copy of the DSP files: # rm -rf /var/crash/dsp If you are working on an unconfigured node, you can now type 'reboot' and press enter to return to the configuration wizard.

GEOM_MIRROR[1]: Adding disk da31p3 to journal-backup. GEOM_MIRROR[1]: Disk da31p3 state changed from NONE to NEW (device journal-backup). GEOM_MIRROR[1]: Device journal-backup: provider da31p3 detected. GEOM_MIRROR[1]: Disk da31p3 state changed from NEW to SYNCHRONIZING (device journal-backup). GEOM_MIRROR[0]: Device journal-backup: rebuilding provider da31p3. GEOM_MIRROR[1]: Disk da31p3 state changed from SYNCHRONIZING to ACTIVE (device journal-backup). GEOM_MIRROR[0]: Device journal-backup: rebuilding provider da31p3 finished. GEOM_MIRROR[1]: Consumer mirror/journal-backup destroyed. GEOM_MIRROR[1]: Device journal-backup: provider da31p3 activated. GEOM_MIRROR[1]: Alerting on device journal-backup: provider da31p3, state 2. GEOM_MIRROR[1]: Disk da29p3 state changed from ACTIVE to DESTROY (device journal-backup). GEOM_MIRROR[0]: Device journal-backup: provider da29p3 destroyed. GEOM_MIRROR[1]: Consumer da29p3 destroyed.