RE: Onbar - failures with pit recovery

Malcolm, There are several OnBar fixes in 9.40.FC5 on your platform, some dealing with shared memory issues and some dealing with PIT restores. For instance: bug_number 165353 description ONBAR WILL REPORT SHMDT ERRNO 22 ON ANY SYSTEMS WHERE A SHARED MEMORY SEGMENT IS SPLIT BY THE OS bug_number 152308 description ONBAR ERRORS WITH SHMAT (22) WHEN SHMVIRTSIZE EXCEEDS 1804288 KBYTES OR WHEN INFORMIX LOGICAL SHM SEGMENT IS MADE OF MANY OS SHM SEGMENTS bug_number 165168 description POINT IN TIME RESTORE AFTER CHANGING GL_DATETIME SETTING FROM STANDARD TO ANY OTHER DOES NOT WORK bug_number 153256 description POINT IN TIME RESTORE DOES NOT WORK IF THERE IS AN OPEN TRANSCATION THAT STARTED BEFORE LOG WHERE LEVEL 0 BACKUP STARTED Without a more detailed error message that's the best I can do. HTH. -Dave David Buchholz Senior Database Administrator Vital Processing Services -----Original Message----- From: owner-informix-list@iiug.org [mailto:owner-informix-list@iiug.org]On Behalf Of malcolm weallans Sent: Wednesday, December 01, 2004 7:15 PM To: informix-list@iiug.org Subject: Onbar - failures with pit recovery We have implemented onbar on IBM AIX 5.2 64 bit with IDS 9.40.fc3. When we do a point-in-time recovery it sticks if fast-recovery for some time and then fails with an assertion failure and problems with shm. I can't give the exact messages at this time as we are trying a point-in-log recovery at the moment and I don't have the ability to dial in when the recovery is underway. For background information we tried three pit recoveries before we went live and two worked. As we had started trying to do a log backup whilst recovering on the attempt that failed we assumed that this was the cause. The site is 24 x 7 and today a junior support person reloaded a table wrongly and corrupted a live table. So we decided to use pit recovery to restore to about 5 minutes before the error occurred. This failed with the assertion failure so we tried another usiing a point-in-time before the current log. That failed as well. I am told that FC5 is now available and fixes problems with similar symptoms to this, but the scheduling of a software upgrade like that is not easy. However, if we were sure that this would fix the recovery I would be able to force installation in a very short time. Can anybody look through the bug lists fixed in FC5 to see if there is anything likely, (FC5 release notes are not on IBM web site yet), or does naybody have any ideas what we are doing wrong. regards Malcolm sending to informix-list