Bug#518431: linux-2.6: NFS locking: task blocked for more than 120 seconds -> NFS client stuck -> many processes wait on D
Package: linux-2.6
Version: 2.6.26-13
Severity: important
Recently I've reported #517122
very slow access/open/... syscalls on NFS mounted files
which got reassigned to linux-2.6.
Now I recall that the node also got the same message from kernel reported
prior it started to have #517122 symptoms which vanished after I restarted
VNC server (restart of original NFS server which caused it at the beginning
didn't actually help iirc).
In 10 days since upgrading to lenny and 2.6.26 (from 2.6.18) I have 2
nodes out of 27 in out cluster stuck -- pts is not accessible so I can't
interactively login, but seems to be be able to run non-pts-needed commands
(e.g. dmesg, etc). Both nodes started to puke those evil messages into their
logs (and netconsoles). brief googling lead me to
NFS regression in 2.6.26?, "task blocked for more than 120 seconds"
http://groups.google.com/group/linux.kernel/browse_thread/thread/d6fb6972a1043d95/1b0978941d189a5c?pli=1
so it seems to be common to the boxes with heavy NFS traffic.
3 patches seems were localized and original reporter reported the
success. It would be great if those are considered for 2.6.26 in Debian.
-- System Information:
Debian Release: 5.0
APT prefers stable
APT policy: (500, 'stable')
Architecture: amd64 (x86_64)
Kernel: Linux 2.6.26-1-amd64 (SMP w/2 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/bash
Reply to: