3.5.1 Matrix Server Release Notes

PolyServe Release Notes 6

it is necessary to reset the timeout, change your scripts to use mx

nfsprobe.

This release includes fixes for the following problems:

• Defect 9167. Under certain conditions, the MxFS-Linux NFS server

could reject a file handle from a client as “stale” even though the file

handle was valid. The conditions related to cached inode management

within the exported PSFS filesystem. The filesystem now properly

handles the inode cache for files accessed through NFS.

• Defect 14832. Some NFS locking clients issue an unlock for all possible

bytes when closing a file, while other clients issue an unlock for each

specific range of locked bytes. Previously, the MxFS-Linux service did

not properly handle the first type of unlock-on-close. MxFS-Linux

now cancels pending lock requests when it receives an unlock for all

possible bytes, even if no pending request exactly matches the unlock

range.

• Defect 15097. File locking operations could hang on NFS clients that

perform strict source checking on their lock reply packets.

• Defect 15191. Under conditions of heavy load, nfsd could incorrectly

close TCP port 2049 and stop responding to all TCP-based requests.

No such issue existed for UDP port 2049.

• Defect 15211. The Export Group device monitor closed an RPC socket

that had already been closed by the operating system, which caused

errors such as “Bad file descriptor,” “RPC: unable to send,” and “RPC:

unable to receive.”

• Defect 15265. Directory read operations over NFS could cause internal

deadlocks in the exported PSFS filesystem. This would affect not only

the NFS clients, but any other processes trying to access the relevant

directory. In some cases, the entire filesystem could be affected. The

locking order has been changed to prevent such deadlocks.

• Defect 15272, 15791. MxFS-Linux probes the NFS service periodically

to monitor its status. Previously, RPC timeouts were treated as service

failures. This allowed temporary resource congestion to trigger

failovers of Virtual NFS Services to other server nodes. RPC timeouts