Whamcloud - gitweb
git://git.whamcloud.com / fs / lustre-release.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 1bd1be8) | patch
b=2230
authorzab <zab>
Wed, 10 Dec 2003 02:03:24 +0000 (02:03 +0000)
committerzab <zab>
Wed, 10 Dec 2003 02:03:24 +0000 (02:03 +0000)
commite018f6948c52757ed7f27628453aa15b7f869adb
tree5981461d68fdb207b986df71f6900ee970af78e2tree | snapshot
parent1bd1be8bee38262465a3c0385a8f4608e17b04f6commit | diff
b=2230
Allocation failures during heavy bulk IO load were causing timeouts.  Using
GFP_NOFS throughout lustre, and in particular instead of 0 as sk->allocation,
is our most recent attempt to appease the VM.  Make lots of noise if you see
allocation failures or deadlocks involving threads waiting for memory.
lnet/include/linux/kp30.h diff | blob | history
lnet/klnds/socklnd/socklnd_cb.c diff | blob | history
lustre/include/liblustre.h diff | blob | history
lustre/include/linux/obd_support.h diff | blob | history
lustre/portals/include/linux/kp30.h diff | blob | history
lustre/portals/knals/socknal/socknal_cb.c diff | blob | history
Lustre primary development and releases