Subject: RE: small memory box, NFS-mounted swapfile panic
To: None <root@garbled.net>
From: Sean Doran <smd@ebone.net>
List: port-alpha
Date: 10/29/1998 08:29:56
Tim Rightnour wrote:

| This is UVM's way of telling you, "I'm out of memory, and out of swap, and Mach
| VM would have just wedged at this point."

I don't think I was out of swap.  Here's the file I was swapping to:

crasse# ls -ls /alpha/swap0
13552 -rw-r--r--  1 root  wheel  134218240 Oct 28 21:04 /alpha/swap0

crasse# df /alpha
Filesystem  1K-blocks     Used    Avail Capacity  Mounted on
/dev/sd0a     1870476   508060  1268892    28%    /alpha

| Adding more swap might stop the panics.. however, it will be slow as hell while
| the pagedaemon starts eating cpu and the disks start running wild.  But I doubt
| this is NFS swap related.. my swap is local, and my panic was the same.
|
| I've only seen this once in 9 or so months with 48meg ram, and 150 mb of swap. 
| And I was doing something particularly nasty at the time. (Other than that,
| I've never had a crash)

I wasn't running much, as you can see from the output of DDB ps in
the previous message.

Moreover, the panic happens predictably when I try to do a make cleandir
in lib/libc.  Intuition suggests that maybe it's attributes-cache related,
but I haven't gone poking around at all to look for something that
needs lots of physical memory and not virtual memory.  The lack of a
stack trace doesn't help. :(

Really slow behaviour would be just fine -- I know the box has no
real memory; panic not so fine.

	Sean.