Search Audit Log - Locks System for Users
Posted: Mon Jun 07, 2010 12:41 pm
Folks
I am just wondering if anyone out there has solved this problem.
When we run a query on the audit log to search for changes, even only a day ago, it takes a considerable time to run. This is not good in itself but this locks out other users while the search is being done.
1. I know our release 9.4 FP1 is problematical in terms of locking so on your system when you search the logs do you get the problem where other users get a lock out.
2. We savedataall every hour which results in a new log file being written from tm1s.log. If we saved less frequently could performance be improved. I assume it quickly searches the portion of the log held in memory and the delay is retreiving and formatting the data in log files held on disk. If we saved less could we improve performance?
3. Is there a tool that searches the logs without the delay. We sometimes use agent ransack but the problem is that we don't have a dimension signature and thus do not have our dimensions in a consistent order in our cubes and hence cannot tell where a particular dimension appears in a log. I did think of writing a loader into sql so that data could be more easily searched but that would take time and effort. Has anyone tried this and is prepared to share experiences.
Once again thank you for your help
Regards
John
I am just wondering if anyone out there has solved this problem.
When we run a query on the audit log to search for changes, even only a day ago, it takes a considerable time to run. This is not good in itself but this locks out other users while the search is being done.
1. I know our release 9.4 FP1 is problematical in terms of locking so on your system when you search the logs do you get the problem where other users get a lock out.
2. We savedataall every hour which results in a new log file being written from tm1s.log. If we saved less frequently could performance be improved. I assume it quickly searches the portion of the log held in memory and the delay is retreiving and formatting the data in log files held on disk. If we saved less could we improve performance?
3. Is there a tool that searches the logs without the delay. We sometimes use agent ransack but the problem is that we don't have a dimension signature and thus do not have our dimensions in a consistent order in our cubes and hence cannot tell where a particular dimension appears in a log. I did think of writing a loader into sql so that data could be more easily searched but that would take time and effort. Has anyone tried this and is prepared to share experiences.
Once again thank you for your help
Regards
John