On Apr 9, 2007, at 11:11 PM, Shane Garoutte wrote:
On Apr 9, 2007, at 5:51 PM, Greg Earle wrote:
Apr 9 17:12:59 notatoaster1 automount[27576]: >> mount: toaster:/ vol/vol1/local failed, reason given by server: Permission denied Apr 9 17:12:59 toaster Mon Apr 9 17:12:59 PDT [nis_main:warning]: Bound to preferred NIS server NNN.NN.NN.69_n Apr 9 17:12:59 toaster Mon Apr 9 17:12:59 PDT [nis_poll:error]: Cannot send NIS binding request to NNN.NN.NN.70 : No buffer space available Apr 9 17:12:59 toaster Mon Apr 9 17:12:59 PDT [nis_poll:error]: Cannot send NIS binding request to NNN.NN.NN.69 : No buffer space available Apr 9 17:12:59 toaster Mon Apr 9 17:12:59 PDT [mountd_main:warning]: yp_match: clnt_call: RPC: Unable to send_ errno = No buffer space available
The way I am reading the above (and its late so I could be completely out of it), this seems to me to be a problem when trying to send a NIS poll to whatever system you are using for NIS. Are you sure your NIS server is happy and has plenty of memory to allocate buffers to your NFS mount requests? Additionally, the way this is formatted (the verbatim wording) reminds me of a Solaris error, you using Sol for NIS?
http://www.adminschoice.com/docs/ solaris_error_message_l_n.htm#errno141
The 2 NIS slave servers (yes, both Solaris 9 boxes) do little else besides service NIS and some application Kerberos and LDAP services. They don't themselves serve out NFS, nor are they much in the way of NFS consumers.
There is only one instance in the logs of issues between the NIS servers and the Filer, i.e. an "NFS server not responding still trying" instance that righted itself about a minute later.
Since I highly doubt they both ran out of buffer space simultaneously, I'm assuming it's the NetApp Filer in question that did. ;)
We've never experienced these "No buffer space available" issues before, and you can tell from the age of the OS that we've had this puppy for quite a while now (and an F540 before that, too).
Someone from the list used the comparison tool to show me that there's literally hundreds upon hundreds of bugs fixed between 6.3.1R1 and 6.5.7. It looks like a couple of them might be relevant to this situation:
43372 SNMP reports "No buffer space available" 4
100694 WAFL ran out of buffers on low memory systems. WAFL 2
I'll plead with the Management to see if there's any way we can get a waiver (we're in a Flight Project Configuration Freeze - people get nervous when you tell them you're going to touch the Filer) to upgrade it.
Thanks for all your suggestions so far - including Nils' one for me to try and get a new NOW account for us. I'm pursuing that route as we speak ...
- Greg