Ask you a question:
A, the question background:
A and B server to install the AIX operating system and make the dual computer hot standby, do HA respectively installed A DB2 database and create the instance, there is A Shared volume group, will create the Shared library volume group, under normal circumstances set mounted on A server, database server database instance of B is started;
Two days before A server outage suddenly, and then the database Shared volume group mount on server B, according to my understanding the database should be the first time to do A crash recovery, query the database on the server B db2diag diagnostic log. The log, and detected the tooltip crash recovery is men, but waited for half A day is unable to connect the database gives person's feeling is crash recovery has yet to be completed (for about 30 minutes), then perform even any library related operation such as library hang live when you execute the command and no response and no error return; Finally also not able to perform db2stop force hang no response, finally carried out only after db2_kill and ipclean in B parked the instance on the server, A server restart to complete at this time, start the database after HA volume group back to A server, A server database instance starts, the database soon crash recovery, the view of A server database diagnostic logs, sent A crash recovery is men information, behind and crash recovery initialization start

Information, and read some LSN, and these information at the time to switch to the database server B when they didn't see the server if no B walk crash recovery process

Two, there is doubt
(1) personally think B no perform crash recovery at all in the database (based on the diagnosis of B server logs to make judgment) at that time, I wonder if this correct?
(2) if I (1) view is right, so why B server did not perform crash recovery? If I (1) view is wrong, so why so long haven't restore successful and immediately back to A server crash recovery is done?
(3) if (1) the view of case, did not start on B server crash recovery, personally think that A and B on the database instances in the start state, namely for A library from A cut on the server, if want to undertake crash recovery, should walk instance startup, loading the database and perform database crash recovery process, at A time when B in the startup state when loading A database instance running on other server database will cause crash recovery has A problem, this is my doubt direction but there is no theoretical support, please give some comments