Subject: Re: le0 timeouts?
To: None <P.D.Helliwell@e-eng.hull.ac.uk>
From: Jacob Suter <jsuter@intrastar.net>
List: port-vax
Date: 12/12/1997 19:24:06
> First of all BIG THANK YOU PEOPLE.  We've finally managed to get a VS3100
> booting from a linux machine and running NetBSD/vax 1,2G :*)  so, we're
> quite happy already.  One thing we have noticed though is that on the VS
> we keep getting "le0 timeout" on the /var/log/messages file.  It was
> running routed when it didn't need to so we killed that and it doesn't do
> it as much.  Now it only tends to do it when there is heavy disk access
> over NFS.  Two theorys spring to mind, either a problem with the VS or a
> problem with the linux 2.0.29 (slackware) NFS server.  I remember a
> problem a while back that some people had with le0 overruns and the 16Mb
> memory thing.  We have 8Mb and it doen't seem to happen.  It's not a huge
> problem as it just glitches for a while and then it's OK again.  Just
> wondered if anyone had seen this.

Don't sweat it - it happens.  It pauses the network for about 3
seconds.  I get it on my 3100m30 w/ 16mb ram connected to a FreeBSD
2.2.5 server.  I asked before and it seems to be "normal".  It happened
to me with both 1.2G and 1.3_alpha, though 1.3_alpha is CONSIDERABLY
less irritating about them (and the network speed 'feels' way faster -
way to go guys!)
 
> After two VS3100s there's a MicroVAX II, then a MicroVAX 3300 to try and
> get running NetBSD.  We're dead chuffed with the VS3100s, it's great to
> see unix running on them :*)  Soon we'll have a small clump of vaxen
> being useful again.

They're suprisingly useful.  I got four 3100's here and they really do
perform well.  I'd like to get something a bit faster (maybe a 3100/M76)
and use it in a real world situation and see how it holds up.  My only
real complaint with NetBSD 1.3_Alpha is the SCSI support on the m30
still being PIO only, but I understand why its still that way...

Well, I was impressed.  I just built the pine/pico package in about 2
hours while supporting a few shells on the box.  All was well until it
wound up >100k/sec of constant swapping over NFS (yuck!) when it was
linking the 'pine' binary - a real memory eater :)

I still want to start that vax-only IRC network...  I really think these
vaxes would rock for medium to large IRC server/networks.

JS