NetApp requires setting a timeout value of 190 seconds for all VMware data stores on their filers to handle “long fabric or storage-side I/O interruptions”; this is done as follows on FreeBSD:
# sysctl kern.cam.da.default_timeout=190
This can be set permanently in /etc/sysctl.conf:
kern.cam.da.default_timeout=190
Elsewhere, I have seen a recommendation for FreeBSD on ESX that kern.cam.da.retry_count be set to 120.
2 Comments
Do this setting prevent FreeBSD from crashing when storage volumes are migrating from a controller to another on the SAN ?
@Milu – Yes, it does keep FreeBSD from crashing during a controller takeover, in my limited experience.
Post a Comment