I'm still at a loss for this issue. Beginning last Wednesday, the Exchange administrator failed the cluster over to another node and all subsequent autosupport tests started working perfectly. I assumed that the other node was the problem, so we again made it the active node on Friday to do more tests. Surprisingly, all autosupport tests and my weekly log email have worked fine (as well as a test today). I don't know if the act of failing the cluster to the other node fixed something, but it's back to working again. It seems this issue only pops up once a month, lasts 3-4 days, then disappears again.
We're having some nasty autosupport issues too. The symptoms are somewhat different from what you described. We were pointed at this:
http://now.netapp.com/NOW/cgi-bin/bol?Type=Detail&Display=263066
The bug report is typically lacking in real information. The bottom line is that once a filer is affected by this bug, the only way out is to change the autosupport.transport to http or smtp, and _REBOOT_. So much for high availability. The scheduled release for a fix is 7.2.5.
Ok...I've read the rest of this thread and I see your filer is generating autosupports and they're getting to NetApp, but not to you. We've been bitten by this in the past, too. First of all, get rid of the Exchange Server. ;) Ok, if you can, point to a Sendmail or other SMTP relay. And check the logs, if they're available to you (gotta love the logs on Exchange...). We found that ours were getting rejected on our MTA because of our self-imposed message-size limit (10MB), which was readily discovered in Sendmail logs. Some of these autosupports are huge. Try setting the autosupport.content option to "minimal" ('options autosupport.content minimal), and generating another test. If you get it, you've found the culprit.
-Kevin
*------------------------------------------*-----------------------* | Kevin Davis (UNIX/Storage Sysadmin) | Natick, Massachusetts | | 508.647.7660 | 01760-2098 | | mailto:kevin.davis@mathworks.com *-----------------------* | http://www.mathworks.com | | *------------------------------------------*-----------------------*