> 2007/8/10, Junko IKEDA <[EMAIL PROTECTED]>: > > Hi, > > > > sfex_lock.c would work like this; > > 1) read lock data > > 2) check current lock status > > 3) reserve lock > > 4) detect the collision of lock > > 5) extension of lock > > 6) lock acquisition completion > > > > in "reserve lock" phase, each node writes its status on the disk, > > and the later node is going to reserve the lock. > > the former one gives up and the race will be end here. > > Assume we have 2 nodes. > 1. Node A & B reach step 3) in the same time. > 2. sfex_lock on Node B is scheduled out due to some other reasons. > 3. sfex_lock on Node A goes through step 3 to 6, and Node A holds the lock now. > 4. sfex_lock on Node B is scheduled back, and goes through step 3 to 6 also.
Node B would not be able to get lock in this case, because Node A has already held the lock. Node B would check the lock status before reserving the lock, and notice Node A has the lock. With this condition, Node B is going to give up to lock. > 5. Now both A & B have the lock. > > > During collision_timeout waiting, the cord around "detect the collision of > > lock" has responsibility to prevent the race. > > in other cases, "check current lock status" would prevent it. > > > > as a precondition to ensure the control exclusively, > > lock_timeout should be longer enough than collision_timeout. > > > > Does that answer you? > > > > Thanks, > > Junko > > > > > > > -----Original Message----- > > > From: [EMAIL PROTECTED] > > > [mailto:[EMAIL PROTECTED] On Behalf Of Xinwei Hu > > > Sent: Thursday, August 09, 2007 8:58 PM > > > To: High-Availability Linux Development List > > > Subject: Re: [Linux-ha-dev] Shared disk file Exclusiveness > > > controlprogramforHB2 > > > > > > 2007/8/9, Junko IKEDA <[EMAIL PROTECTED]>: > > > > Hi, > > > > > > > > sorry, my previous answer was off the mark... > > > > When 2 nodes reach there at the same time, > > > > node A notices that the other node want to lock too, so give up lock > > itself. > > > > > > I only see that you sleep a period of collision_timeout. This will not > > prevent > > > race condition from happening. > > > Am i missing anything else ? > > > > > > > node B is ready to lock. > > > > > > > > Thanks, > > > > Junko > > > > > > > > > > > > > -----Original Message----- > > > > > From: [EMAIL PROTECTED] > > > > > [mailto:[EMAIL PROTECTED] On Behalf Of Junko > > IKEDA > > > > > Sent: Thursday, August 09, 2007 4:30 PM > > > > > To: 'High-Availability Linux Development List' > > > > > Subject: RE: [Linux-ha-dev] Shared disk file Exclusiveness control > > > > > programforHB2 > > > > > > > > > > Hi, > > > > > > > > > > You know that could be true... > > > > > but if it's called from RA, 2 nodes wouldn't reach that part at the > > same > > > > > time, right? > > > > > One node will be able to reach there according to the score rule. > > > > > > > > > > Thanks, > > > > > Junko > > > > > > > > > > > -----Original Message----- > > > > > > From: [EMAIL PROTECTED] > > > > > > [mailto:[EMAIL PROTECTED] On Behalf Of Xinwei > > Hu > > > > > > Sent: Thursday, August 09, 2007 3:00 PM > > > > > > To: High-Availability Linux Development List > > > > > > Subject: Re: [Linux-ha-dev] Shared disk file Exclusiveness control > > > > program > > > > > > forHB2 > > > > > > > > > > > > Hi, > > > > > > > > > > > > There's races in the sfex_lock.c. > > > > > > > > > > > > When 2 nodes reach sfex_lock.c:265 in the same time. > > > > > > > > > > > > node A: reserve lock -> wait for collision_timeout -> hold lock > > > > > > node B: sleep -> sleep -> > > > > reserve > > > > > > lock > > > > > > > > > > > > 在 星期三 08 八月 2007 12:00，NAKAHIRA Kazutomo 写道： > > > > > > > Hello, all. > > > > > > > > > > > > > > We wrote a Shared Disk File EXclusiveness Control Program, called > > > > > > > "SF-EX" for short, could prevent a destruction of data on > > > > > > > shared disk file system due to Split-Brain. > > > > > > > > > > > > > > This program consists of CUI commands written in the C and RA, > > > > > > > the former is used for managing and monitoring shared disk status > > > > > > > and the latter is the same as other common RAs. > > > > > > > > > > > > > > We tested this program on IBM and HP platform, and we confirmed > > > > > > > all functions worked well. > > > > > > > > > > > > > > Our test environment is listed below: > > > > > > > Software: > > > > > > > OS :RHEL4 ES Update5(kernel2.6.9-55.ELsmp) > > > > > > > HB2:heartbeat-2.1.2-2 > > > > > > > > > > > > > > Hardware: > > > > > > > IBM platform: > > > > > > > Server : System x3650 > > > > > > > Shared disk: DS 4700 (FC) > > > > > > > > > > > > > > HP platform: > > > > > > > Server : DL380G5 > > > > > > > Shared disk: MSA500G2 (SCSI) > > > > > > > > > > > > > > How to install and configuration are described in README. > > > > > > > If you are interested in this program, please try it and > > > > > > > let me know your comments. > > > > > > > > > > > > > > Your suggestions on how to improve are really appreciated. > > > > > > > > > > > > > > Best regards. > > > > > > _______________________________________________________ > > > > > > Linux-HA-Dev: [email protected] > > > > > > http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev > > > > > > Home Page: http://linux-ha.org/ > > > > > > > > > > _______________________________________________________ > > > > > Linux-HA-Dev: [email protected] > > > > > http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev > > > > > Home Page: http://linux-ha.org/ > > > > > > > > _______________________________________________________ > > > > Linux-HA-Dev: [email protected] > > > > http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev > > > > Home Page: http://linux-ha.org/ > > > > > > > > _______________________________________________________ > > Linux-HA-Dev: [email protected] > > http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev > > Home Page: http://linux-ha.org/ > > _______________________________________________________ Linux-HA-Dev: [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha-dev Home Page: http://linux-ha.org/
