[gpfsug-discuss] Node expulsion from GPFS Cluster

[Jonathan Monk]

We saw something similar with disk leases that was due to load on the manager node but that was with a much simpler config than you describe. Do you have a sample extract from mmfs.log ?

Jonathan
________________________________________
From: gpfsug-discuss-bounces at gpfsug.org on behalf of Linda Dewar
Sent: Wednesday, January 16, 2013 4:00:15 PM
To: gpfsug-discuss at gpfsug.org
Subject: [gpfsug-discuss] Node expulsion from GPFS Cluster

Hello,

Just wondering if anyone else has seen anything like this.

I am using GPFS  3.4.0.7.

I have a multicluster setup with NSD servers and TSM servers (running
RHEL 6.2) in a Server cluster and clients (some running SUSE 11 SP1 and
some running RHEL 6.2) in a Client cluster.

All clients in both clusters are connected to a IBM BNT G8200 switch

Nodes in the Client cluster are regularly expelled from the cluster,
either  at the request of other nodes in the Client cluster, or nodes in
the Server cluster. This means that GPFS is shut down on the expelled
node and filesystems unmounted . Obviously inconvenient to the users.

Basic connectivity (ie response to ping) between nodes is unaffected at
the times of the node expulsions. We are looking at using Wireshark to
investigate what appears to be some sort of network connectivity problem.

Does any of this sound familiar to anyone else, and does anyone have any
suggestions as to what we should be looking for?


Many thanks,

Linda

--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.

_______________________________________________
gpfsug-discuss mailing list
gpfsug-discuss at gpfsug.org
http://gpfsug.org/mailman/listinfo/gpfsug-discuss


The University of Dundee is a registered Scottish Charity, No: SC015096