full debug log from arla lockup/non-responsiveness

[Neulinger, Nathan R.]

I set both the xfs and arla debugging to all. 

The lockup seems to affect particular directories. Also, if the directory
gets in this state (where it locks up) a flushv on the directory will
completely freeze the system  to where it will answer pings but nothing
else.

The full debug log is here:

http://www.umr.edu/~nneul/debug-traces/arla-webindex-19990211-ls

The first is the output from 'ls directory' then waiting a while, then
control-C. 

The flushv only got the following before locking up the machine (no oops):

	Feb 11 10:59:15 webindex arla[361]: probe (192.65.97.1)
	Feb 11 10:59:20 webindex arla[361]: worker 0: processing
	Feb 11 10:59:20 webindex arla[361]: Rec message: opcode = 22
(pioctl), size = 2096
	Feb 11 10:59:20 webindex arla[361]: sending wakeup: seq = 1273148,
error = 22

The arla config is:

	high_vnodes 4000
	low_vnodes 3000
	numcreds 100
	numconns 100
	numvols 100
	fpriority 100
	high_bytes 150M
	low_bytes 140M

The cache status shortly after the time of the hangup:

	Arla is using 2929 of the cache's available 153600 1K byte blocks
	(and 805 of the cache's available 4000 vnodes)

If there is anything else you want me to get?

-- Nathan

------------------------------------------------------------
Nathan Neulinger                       EMail:  nneul at umr.edu
University of Missouri - Rolla         Phone: (573) 341-4841
Computing Services                       Fax: (573) 341-4216