em0 freezes on ZFS server

[Willem Jan Withagen]

On 25-2-2010 23:59, Jack Vogel wrote:
> The failure to "setup receive structures" means it did not have sufficient
> mbufs
> to setup the RX ring and buffer structs. Not sure why this results in a
> lockup,
> but try and increase kern.ipc.nmbclusters.
>
> Let me know what happens,

I've doubled the value 25600 => 51200.

This is wat netstat -m told me when it refused to revive em0:

24980/2087/27067 mbufs in use (current/cache/total)
24530/1070/25600/25600 mbuf clusters in use (current/cache/total/max)
22217/741 mbuf+clusters out of packet secondary zone in use (current/cache)
0/35/35/12800 4k (page size) jumbo clusters in use (current/cache/total/max)
0/0/0/6400 9k jumbo clusters in use (current/cache/total/max)
0/0/0/3200 16k jumbo clusters in use (current/cache/total/max)
55305K/2801K/58106K bytes allocated to network (current/cache/total)
0/5970/2983 requests for mbufs denied (mbufs/clusters/mbuf+clusters)
0/0/0 requests for jumbo clusters denied (4k/9k/16k)
0/0/0 sfbufs in use (current/peak/max)
0 requests for sfbufs denied
0 requests for sfbufs delayed
1011716 requests for I/O initiated by sendfile
0 calls to protocol drain routines

Now I've seen some discussion on the list suggesting that full mbuf could 
also be because the device is down and the queue builds up rather rappidly 
in the mbufs.

Probably the reason why this happened yesterday is that I started doing 
major software builds (over ZFS/NFS/TCP/v3) against data stored on this box.

--WjW