Hi folks,
I created a patch against Release 1.3.2 to fix a bug in event logging
with mpe. It has been originally found and fixed by Frank Panse here in
Heidelberg.
The problem is, that MPE2 requires stop event to have the following
number of start event of the same type, otherwise clog2TOslog2 (f
These patches fix problems that can crop up if a server is shutdown
while it is still trying to service I/O operations. The first one
depends on perf counter patches from
http://www.beowulf-underground.org/pipermail/pvfs2-developers/2005-December/001704.html,
but only because of a minor change
perf-counter-client.patch:
--
This patch adds performance counter support to pvfs2-client:
- new state machines to periodically rollover the values
- instrumentation of the acache
- command line (to pvfs2-client) as well as /proc/sys/pvfs2 hooks to
modify pe
These patches largely depend on the acache patches from this posting:
http://www.beowulf-underground.org/pipermail/pvfs2-developers/2005-November/001606.html
perf-counter-api.patch:
-
This patch overhauls the performance counter api to add quite a bit of
funct
msgpair-gossip-svr.patch:
--
This is a small patch that just changes the error message in
msgpairarray that gets printed if an operation fails. It attempts to
print the name of the server that communication fails to if possible,
and gives a little bit more indication of
pvfs2-kernel-op-timeout.patch
--
This is a patch that changes the default operation timeout in the kernel
from a hard coded 60 seconds to a dynamic value that can be tuned either
via insmod options or vi a /proc/sys/pvfs2 file. If nothing else, this
is handy for deb
Most of these are fairly straightforward:
pvfs2-kernel-revalidate.patch:
--
This removes a call to make_bad_inode() that occurs in the vfs
d_revalidate path if a getattr fails. make_bad_inode() isn't necessary
for error propigation here, and was causing the inode c