LSI 3108 Watchdog Timeout Causing Reboot

[TheGagne]

My ZFS system is crashing about once per month due to Watchdog Timeout. Not ideal. Also due to the irregularity of the crashes, it is hard to troubleshoot. I spent a bunch of time trying out how to disable watchdog reboots, but that did not solve the issue (ipmitool mc watchdog get, ipmitool mc watchdog off).

# FreeNAS-9.3-STABLE-201512121950

# from /var/log/messages before system reboot due to watchdog timeout
Aug 8 09:37:41 freenas mfi0: COMMAND 0xfffffe0001330550 TIMEOUT AFTER 975 SECONDS

# I'm using lsi3108 in jbod mode

Code:

[root@freenas] ~# mfiutil show adapter
mfi0 Adapter:
    Product Name: LSI 3108 MegaRAID
   Serial Number:
        Firmware: 24.7.0-0026
     RAID Levels: JBOD, RAID0, RAID1, RAID5, RAID6, RAID10, RAID50
  Battery Backup: not present
           NVRAM: 32K
  Onboard Memory: 2048M
  Minimum Stripe: 64k
  Maximum Stripe: 1M

# some posts of others with these issues point to changing read ahead and drive caching settings as a potential fix.
# https://forums.freenas.org/index.php?threads/lsi-megaraid-sas-9261-8i-timeout-with-write-back.5960/
# suggests drive write cache: disabled (what is default?)
# https://forums.freebsd.org/threads/20658/
# suggests read ahead: adaptive
# neither of those are using 3108 though...

# here's my settings on mfid0

Code:

[root@freenas] ~# mfiutil cache mfid0
mfi0 volume mfid0 cache settings:
             I/O caching: disabled
           write caching: write-through
write cache with bad BBU: disabled
              read ahead: adaptive
       drive write cache: default

# other thoughts are possibly to update firmware on lsi card or update driver.
# I understand my use of 3108 is non optimal - if I purchase an lsi 9300 12gb hba, will I be able to simply import my pool or will it require drive reformat since they are raid0 formatted?

 

[dlavigne]

Were you able to resolve this?

 

[TheGagne]

No, not yet (although it hasn't crashed since posting -- it crashes roughly once per month).

My goal is to #1 purchase a full backup system for this (I have about 75% backed up now), and #2 buy an HBA to replace the raid card.