[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: NFS disconnects periodically



Am Mittwoch, 8. Dezember 2004 13:07 schrieb Ralf Gesel|ensetter:
> Thanks for any further hints on how to corner the bug.

Hi there,

things are getting worse - but maybe this was a final outage - and now I 
conquered (there is still hope!) :)

I tried setting all NICs to 100 FD by means of ethtool.
At some point, it has been ltspserver refusing to interact via eth1 
which is the one connected to the backbone. Turning autoneg on wakened 
the thin clients (at one situation).

I now started nload for all NICs concerned:
tjener:eth0 (10.0.2.2)
ltsp:eth1 (10.0.2.10)
ltsp:eth0 (192.168.0.254)

While both tjener:eth0 and ltsp:eth0 delivered rates up to 50 Mbit and 
more, ltsp:eth1 _could_ write to tjener (outgoing) with nearly 100 Mbit 
- but the other way around kept always below 5 Mbit - on average rather 
100 kbit. I had similar observations when copying files between tjener 
and a workstation.

When the whole windows room is surfing, tjener:eth0 goes up very fast. 
So it seems like it would be the route between tjener ---> ltsp that is 
too slow. A switch matter? Or an question of services (NFS / Ldap)? 

We already had been surprised that creating some 1000 accounts had taken 
us 8 hours, and when I enter the login name at a thin client, it takes 
now at least a second until the default session is determined. Is there 
a way in monitoring ldap's / NFS's response time?

Very tired now
Ralf



Reply to: