Is there maybe some IO problem on the iSCSI target side?
IIUIC the problem is some timeout, which could indicate that the target
is overloaded.
Maybe. I need to check with Dell. I did manage to get it to be a little
more stable with this config.
defaults {
polling_interval 10
path_selector "round-robin 0"
path_grouping_policy multibus
getuid_callout "/usr/lib/udev/scsi_id --whitelisted
--replace-whitespace --device=/dev/%n"
path_checker readsector0
rr_min_io_rq 100
max_fds 8192
rr_weight priorities
failback immediate
no_path_retry fail
user_friendly_names no
}
devices {
device {
vendor COMPELNT
product "Compellent Vol"
path_checker tur
no_path_retry fail
}
}
I referenced it from
http://en.community.dell.com/techcenter/enterprise-solutions/w/oracle_sol....
I modified it a bit since that is Red Hat 5 specific and there have been
some changes.
It's not crashing anymore but I'm still seeing storage warnings in
engine.log. I'm going to be enabling jumbo frames and talking with Dell
to figure out if it's something on the Compellent side. I'll update here
once I find something out.
Thanks again for all the help.