MC/ServiceGuard NFS Toolkit Version A.11.00.03 Release Notes

Limitations

1. NFS/TCP may not be able to service new requests if the default kernel parameter

max_thread_proc is hit.

Users may need to increase the value of the kernel parameter, max_thread_proc in

/stand/system on the HA server machine(s). They should then rebuild/reboot the kernel to

take advantage of the change. The setting range for max_thread_proc is:

64 <= max_thread_proc <=nkthread

50 <= nkthread <=30000

2. Performance degradation may occur if manual switching between primary node and

adoptive node is done immediately after a failover recovery.

While manually moving NFS package back and forth between primary node and adoptive

node, users must wait at least 10 minutes after the recovery, so that TCP connections get

cleaned up. Otherwise they will see the client applications hanging or taking a long time to

ﬁnish.

Known Problems and Workarounds

File locks are not preserved when MC/ServiceGuard NFS package moves over to a new node.

Any applications that use ﬁle locking must reclaim their locks after an NFS server package

fails over.

MC/ServiceGuard NFS over NFS/TCP cannot be used for an NFS diskless cluster server.

Compatibility and Installation Requirements

For instructions on installing and conﬁguring MC/ServiceGuard NFS Toolkit, see the manual

Managing MC/ServiceGuard NFS (HP part number B5140-90005).