3.5.1 Matrix Server Release Notes

PolyServe Release Notes 7

during NFS service probes are no longer treated as service failures and

will not trigger failovers.

• Defect 15322. The ExportFSsync process could dump core during

product shutdown. This defect did not adversely affect the availability

of NFS exports.

• Defect 15360. More messages from vstatd (the MxFS-Linux version of

rpc.statd) are now sent to syslog rather than discarded. This will help

diagnose startup problems for vstatd.

• Defect 15369. MxFS-Linux now monitors a wider variety of conditions

necessary for successful access to the NFS service to determine service

status. The mount and status RPC protocols are probed periodically,

and the number of running NFSD threads is compared to the number

configured to run.

• Defect 15432. Under certain conditions involving NFS lock requests

that coincide with the byte range of a previously granted lock, MxFS-

Linux could cause the VFS layer to lose track of the granted lock,

while the PSFS filesystem layer still held the lock. This could prevent

an affected client from reclaiming its lock after a virtual host failover

to a new server node. This affected only the Red Hat implementation

of MxFS-Linux. The VFS and PSFS layers now remain in agreement

about granted lock ranges.

• Defect 15823. The default number of NFS Server threads and the

maximum number of outstanding

NFS RPC requests for the TCP and

UDP network transports needed to be increased. As described earlier,

these values are now set when MxFS-Linux is installed.

• Defect 15888. NFS client applications could leave a lock unavailable

forever.

• Defect 16077. When creating a file or directory in a PSFS filesystem,

the MxFS-Linux NFS server could return a different file handle than

when looking up an existing file or directory. This caused Linux NFS

clients to create two proxy files, with independently maintained

attributes and content caches. This manifested in a variety of ways,

often appearing to implicate page cache management in the client. The

problem was observed most often when transitioning a Virtual NFS