Whamcloud - gitweb
LU-1520 ldlm: improve ldlm_pools_shrink algorithm 70/3270/2
authorHongchao Zhang <hongchao.zhang@whamcloud.com>
Tue, 4 Sep 2012 15:27:09 +0000 (23:27 +0800)
committerHongchao Zhang <hongchao.zhang@whamcloud.com>
Tue, 4 Sep 2012 15:27:09 +0000 (23:27 +0800)
commitc861cc7e0b6f7e82fd55b9658dd29578f97b5607
tree6669bbf6e0b0b8a11bccc582b79ed2e3e58959b7
parentbd649b14d6895ce808047ce0edda20928cca0dc1
LU-1520 ldlm: improve ldlm_pools_shrink algorithm

1, shrink namespaces by batches of 64 namespaces, the batch is
   implemented as list
2, limit number of simultaneously shrinking threads to 32 threads
3, have ldlm_pools_recalc to operate with namespaces similar to
   ldlm_pools_shrink
4, use glboal counters of unused locks on client and granted
   locks on servers to avoid iterating over namespaces

Change-Id: I5fb3f56748ae10961c50b4b06c300c9c7f5fca87
Signed-off-by: Vladimir Saveliev <valdimir.saveliev@oracle.com>
Signed-off-by: Hongchao Zhang <hongchao.zhang@whamcloud.com>
lustre/include/lustre_dlm.h
lustre/ldlm/ldlm_internal.h
lustre/ldlm/ldlm_lock.c
lustre/ldlm/ldlm_lockd.c
lustre/ldlm/ldlm_pool.c
lustre/ldlm/ldlm_resource.c