I have a question about the timing on upgrading disk firmware. With the recent DOT's when you do a DOT upgrade, after you reboot and a disk firmware upgrade is needed, DOT warns you at boot time, at which point you can do a disk_fw_update. This is when I've upgraded disk firmware. However, the NOW documentation isn't clear if this is the appropriate time to upgrade the disk firmare.
So, my question is, when is the best time to upgrade the disk firmware:
1) After you've run install_netapp, but before you've run 'download'. 2) After you've run download. 3) After you've rebooted on the new DOT version.
Also, are there any cluster considerations associated with the choice of timing? I noticed when upgrading a cluster this weekend that I received downrev messages from the NetApps until both partners had the same disk firmware.
j. -- Jay Soffian jay@cimedia.com UNIX Systems Engineer 404.572.1941 Cox Interactive Media
Jay,
I just finished this task Sunday with the help of our rep. Our procedure was to push DOT to the filer, run the download and then reboot the filer. When the filer is booting, it will notify you that n disks are out of rev. This is not fatal, just a warning. After the filer was running we performed the disk_fw_update.
Yes, there is cluster considerations. 1.) disable clustering. 2.) down the partner, halt -f. 3.) run the disk_fw_update. 4.) halt -f the system and boot the partner. 5.) run the disk_fw_update.
We disabled the clustering on the pair and performed the disk_fw_update on each filer. When we attempted to reboot the machine we received a error about the mailbox being inconsistent. This was easily fixed when we powered off the disks and the unit head and brought everything back up from a cold restart. You will incur downtime with the upgrade but I understand that engineering knows about this already and are working on it.
-gdg
Jay Soffian wrote:
I have a question about the timing on upgrading disk firmware. With the recent DOT's when you do a DOT upgrade, after you reboot and a disk firmware upgrade is needed, DOT warns you at boot time, at which point you can do a disk_fw_update. This is when I've upgraded disk firmware. However, the NOW documentation isn't clear if this is the appropriate time to upgrade the disk firmare.
So, my question is, when is the best time to upgrade the disk firmware:
- After you've run install_netapp, but before you've run 'download'.
- After you've run download.
- After you've rebooted on the new DOT version.
Also, are there any cluster considerations associated with the choice of timing? I noticed when upgrading a cluster this weekend that I received downrev messages from the NetApps until both partners had the same disk firmware.
j.
Jay Soffian jay@cimedia.com UNIX Systems Engineer 404.572.1941 Cox Interactive Media
"G" == G D Geen geen@msp.sc.ti.com writes:
G> Jay, I just finished this task Sunday with the help of our rep. G> Our procedure was to push DOT to the filer, run the download G> and then reboot the filer. When the filer is booting, it will G> notify you that n disks are out of rev. This is not fatal, G> just a warning. After the filer was running we performed the G> disk_fw_update.
G> Yes, there is cluster considerations. 1.) disable clustering. 2.) down the partner, halt -f. 3.) run the disk_fw_update. 4.) halt -f the system and boot the partner. 5.) run the ddisk_fw_update.
G> We disabled the clustering on the pair and performed the G> disk_fw_update on each filer. When we attempted to reboot the G> machine we received a error about the mailbox being G> inconsistent. This was easily fixed when we powered off the G> disks and the unit head and brought everything back up from a G> cold restart. You will incur downtime with the upgrade but I G> understand that engineering knows about this already and are G> working on it.
This is exactly the procedure I just followed this weekend. However, I didn't have to power-cycle anything. The exact procedure I followed w/o incident was:
1) halt filer A, allow filer B to takeover. 2) upgrade firmware on filer A (had to upgrade from 2.1_a2 to 2.2_a2). 3) give back from filer B. 4) halt filer B, allow filer A to takeover. 5) upgrade firmware on filer B (had to upgrade from 2.1_a2 to 2.2_a2). 6) give back from filer A. 7) disable clustering 8) run install_netapp on filer A. 9) download on filer A. 10) reboot filer A. 11) run disk_fw_update on filer A. 12) reboot filer A (for good measure). Received notification about 'downrev' which was because (I assume) filer B still had older disk firmware. 13) run install_netapp on filer B. 14) download on filer B. 15) reboot filer B. 16) run disk_fw_update on filer B. 17) reboot filer B (for good measure). 18) reboot filer A (for good measure). 19) enable clustering.
This procedure worked fine for me. It is also the procedure I believed to be correct after reading the NetApp online documentation. It was only in hindsight that I thought to ask on toasters about the timing of the disk_fw_update. I am surprised to find about possibly needing to power-cycle the disk shelves. Is this a bug? If so, a search for disk_fw_update on NOW doesn't turn up any documentation on this bug, nor can I find this documented elsewhere on NOW.
j. -- Jay Soffian jay@cimedia.com UNIX Systems Engineer 404.572.1941 Cox Interactive Media
-
G D Geen -
Jay Soffian