MC/ServiceGuard NFS Toolkit Versions A.11.00.04 and A.11.11.01 Release Notes

MC/ServiceGuard NFS Toolkit Versions A.11.00.04 and A.11.11.01 Release Note
Known Problems and Workarounds
Chapter4
Known Problems and Workarounds
The following is a list of the known problems and workarounds in MC/ServiceGuard NFS
Toolkit A.11.00.04 and A.11.11.01. Each of the following describes a problem and the
workaround for each problem. All of the following apply to MC/ServiceGuard NFS Toolkit
A.11.00.04 and A.11.11.01
• File locks are not maintained when an NFS server package moves to an adoptive server.
Any applications that use file locking must reclaim their locks after an NFS server
package fails over. An application that loses its file lock as a result of an NFS package
failover will not be notified. If the server is also an NFS client, any file locks it holds will be
lost.
• An NFS/TCP client operation receives “NFS server not responding still trying” message
while attempting to access the server although the server is up. In this situation, users can
increase the default kernel parameter max_thread_proc by using SAM on HA servers.
• Performance degradation may occur if manual switching between primary server and
adoptive server is done immediately after a failover recovery. After manually moving the
NFS package between the primary server and adoptive server, wait at least 10 minutes
after the recovery while TCP connections clean, otherwise the client applications hang or
take a long time to finish.
• NFS package exported file systems on HA servers are not exported if you restart NFS
clients or NFS servers. Manually switch the NFS package to a different HA server before
stopping and restarting NFS clients and servers with nfs.client stop/start or
nfs.server stop/start.
• HP recommends not enabling AutoFS on the HA server. Enabling AutoFS on the HA
server creates a conflict between AutoFS’s design and the HA solution. Halting NFS
packages may fail if actions to stop and restart NFS clients and servers are taken before
halting NFS packages. This failure can also occur when restarting the AutoFS daemon on
the HA server. If you must enable AutoFS on the HA server, manually switch the NFS
package to a different HA server before restarting the AutoFS daemon or before
performing any stopping or restarting actions on NFS clients and servers.