On Wed, 10 Jan 2007 10:32:54 -0500, Roland P. Chung <[EMAIL PROTECTED]> wrote:
>To all DB2/VM server users: several instances of DB2/VM server abends wi th ARIYI36 05 >have been reported to IBM lately when users migrated their DB2/VM server s to z/VM 5.20 running in a z9 hardware with VMDSS enabled. > > VM APAR VM63968 has been opened to address this problem. > > IBM z/VM development is currently, with the assistance of DB2 for VSE& amp;VM Team, testing the fix for VM APAR VM63968. > > DB2/VM server's ARIYI36 05 abend is an "after the fact" situation. Tha t means, DB2/VM server detects an "inconsistance" between the BDISK and the DBEXTENT when it requests CP to bring in a datapage from the DBEXTENT. Wi th VMDSS support enabled, CP is responsible to move datapage between DASD(DBEXTENT) and dataspace then DB2/VM server moves the datapage from dataspace to its buffer for processing.The error could have been: > 1) CP returns an incorrect datapage > 2) the datapage returned has been previously stored on DASD incorrectl y by CP > > > If a user would like to migrate DB2/VM server to z/VM 5.20 running in a z9 hardware BEFORE the official PTF for APAR VM63968 is available, the recommendation is: > > 1) apply all available MCLs to z9 box; > 2) apply all current VM APARs/PTFs, especially VM63978/UM31822; > 3) disable VMDSS support in DB2/VM server; > 4) if not possible to disable VMDSS support in DB2/VM server, the DB2/ VM server should be archived more frequency. > > >With best regards, > >...Roland Chung >Senior Technical Specialist (S/390,VM/VSE,DB2/VSE&VM) >MAXC Consultants Inc. >Voice/Fax: 416-901-9442 (If busy, call: 416-469-2268) >197 Hastings Ave., Toronto, Ontario, Canada. M4L 2L6 > >** Life is short. Stop once in a while and smell the roses. ** Thanks, Roland. A slight correction - VM63968 is not intended to address any DB2 server abends, it only addresses server / system hangs and hopefully some of the performance problems seen with the VM63968 circumvention with which many z/VM 5.2.0 DB2 customers are running. The VM63968 circumvention is available to any customer that needs it, and it reduces the hang exposure to effectively zero, but can potentially impact system performance negativel y. The latest version of the VM63968 fix itself is just about ready for fixtest, the PTF will hopefully be generally available within the next mo nth or so. The known fixes for DB2 server abends are VM63978 (available since June 2006) and now newly opened APAR VM64120, which is not yet closed, but for which a fixtest is available. z/VM 5.2.0 DB2 customers should certainly be advised to apply VM63978 if they haven't already (regardless of hardware level), and VM64120 when it becomes available (which should not be much longer). There are still some open questions whether there are any DB2 functional exposures even with these two fixes applied, and we are active ly working on trying to answer those questions. Regards, - Bill Holder, z/VM Development and Service, IBM Endicott
