Anne Moore writes:
> My two nics are setup for "Standby" in a group. They keep failover over to
> each other, causing us massive down time.
I'd suggest [EMAIL PROTECTED] as a more likely place
to look for help. "opensolaris-discuss" is a general-interest list,
and as such has a *lot* of traffic. Many of the experts in this area
don't bother subscribing to opensolaris-discuss because the S/N ratio
is too low for them. :-/
> Aug 11 16:28:28 MyOpenSolarisBox in.mpathd[108]: [ID 832587 daemon.error]
> Successfully failed over from NIC ce2 to NIC ce1
> Aug 11 16:29:34 MyOpenSolarisBox in.mpathd[108]: [ID 168056 daemon.error]
> All Interfaces in group production have failed
This looks like the crucial bit. What happened that (a) caused the
system to think that ce1 had recovered and (b) then caused it to think
that it had failed again?
If you're using probe-based failure detection, then one possibility
would be that your probe target is intermittently failing to respond
to ICMP Echo ("ping") messages. When it responds, the system
recovers. When it fails, the system attributes that as interface
failure.
> Any ideas of where, and what tools I can use to determine the issue with the
> Ce1 interface?
Here are a few:
- Make sure you're logging all error messages with "daemon.debug" in
your /etc/syslog.conf file.
- Use "snoop ce1 icmp" to check for those ICMP Echo messages and
find out what's happening to them.
- Use "route monitor" to watch the interface flags. It's possible
that the problem isn't ICMP Echo at all, but rather that the
interface RUNNING flag is flapping -- possibly caused by a bad
cable or switch.
- Kill off the /usr/lib/inet/in.mpathd process and restart it
manually with the "-d" flag to enable debug. This will print all
of its activity in excruciating detail on standard output.
--
James Carlson, Solaris Networking <[EMAIL PROTECTED]>
Sun Microsystems / 35 Network Drive 71.232W Vox +1 781 442 2084
MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677
_______________________________________________
opensolaris-discuss mailing list
[email protected]
