[Drbd-dev] DRBD8: unexpected cstate (PausedSyncT) in receive_bitmap

[Montrose, Ernest]

Hi all,
We are seeing a problem where during syncing the peer_isp bit ends up
getting set when it should not.
I cannot duplicate it but the log clearly shows why the problem
occurred.  Essentially, we never transitionsed
from PausedSyncS -> SyncSource as the volume resumed from a suspended
state.  But yet on the sync source we show
a completion of the sync.  The failure message then occurred on the sync
target.  Below is an *illustrated* version of the
log.  I hope I don't confuse things more:
 
-------------------------SyncSource
side--------------------------------------------------------------------
-------
Mar 16 08:33:16 chobs kernel: drbd1: aftr_isp( 1 -> 0 )
Mar 16 08:33:16 chobs kernel: drbd1: conn( PausedSyncS -> SyncSource )
peer_isp( 1 -> 0 )
Mar 16 08:33:16 chobs kernel: drbd1: Syncer continues.
Mar 16 08:33:16 chobs kernel: drbd1: conn( SyncSource -> PausedSyncS )
aftr_isp( 0 -> 1 )
Mar 16 08:33:16 chobs kernel: drbd1: Resync suspended
Mar 16 08:33:16 chobs kernel: drbd1: peer_isp( 0 -> 1 )
Mar 16 08:33:16 chobs kernel: drbd1: aftr_isp( 1 -> 0 ) <<<<#####Should
have been peer_isp I think ######
##############Under normal operation the two lines would be seen But
they are NOT################# 
#drbd1: conn( PausedSyncS -> SyncSource ) aftr_isp( 1 -> 0 )
#drbd1: Syncer continues.
############################### instead continue below ############
Mar 16 08:33:16 chobs kernel: drbd1: Resync done (total 1 sec; paused 0
sec; 356 K/sec) #### Are we really done here???
Mar 16 08:33:16 chobs kernel: drbd1:  uuid[History_start] now
FC98CEA3D33591C1
Mar 16 08:33:16 chobs kernel: drbd1:  uuid[Bitmap] now 0000000000000000
Mar 16 08:33:16 chobs kernel: drbd1: conn( PausedSyncS -> Connected )
pdsk( Inconsistent -> UpToDate ) ####!WE ARE GOING TO Conneceted from
PausedSyncS..So Same state!!!
#######!!!!!!! ABOVE state should have been conn( SyncSource ->
Connected ) pdsk( Inconsistent -> UpToDate )!!!!!
Mar 16 08:33:16 chobs kernel: drbd1: Writing meta data super block now.
Mar 16 08:33:16 chobs kernel: drbd1: drbd_sync_handshake:
Mar 16 08:33:16 chobs kernel: drbd1: self
269C54A410CF9CAB:0000000000000000:FC98CEA3D33591C1:2EFBE8CDF29C591F
Mar 16 08:33:16 chobs kernel: drbd1: peer
FC98CEA3D33591C0:0000000000000000:2EFBE8CDF29C591E:40519581F499C446
Mar 16 08:33:16 chobs kernel: drbd1: uuid_compare()=2 by rule 8

------------------------------Sync Target
side--------------------------------------------------
Mar 16 08:33:17 chobp kernel: drbd1: peer_isp( 1 -> 0 )
Mar 16 08:33:17 chobp kernel: drbd1: conn( PausedSyncT -> SyncTarget )
aftr_isp( 1 -> 0 )
Mar 16 08:33:17 chobp kernel: drbd1: Syncer continues.
Mar 16 08:33:17 chobp kernel: drbd1: conn( SyncTarget -> PausedSyncT )
aftr_isp( 0 -> 1 )  ### should have been peer_isp I think!!!
Mar 16 08:33:17 chobp kernel: drbd1: Resync suspended
Mar 16 08:33:17 chobp kernel: drbd1: peer_isp( 0 -> 1 ) <<<#### should
have been aftr_isp ########
Mar 16 08:33:17 chobp kernel: drbd1: peer_isp( 1 -> 0 ) <<<#### should
have been aftr_isp ########
#######The 3 lines below would be printed BUT DID NOT##############
#drbd1: conn( PausedSyncT -> SyncTarget ) peer_isp( 1 -> 0 )
#drbd1: Syncer continues.
#drbd1: Resync done (total 1 sec; paused 0 sec; 356 K/sec)
#################Instead got##############
Mar 16 08:33:17 chobp kernel: drbd1: unexpected cstate (PausedSyncT) in
receive_bitmap
 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.linbit.com/pipermail/drbd-dev/attachments/20070328/f862757d/attachment.htm