Slow client load (OpenInsight 64-bit)
At 17 MAY 2022 03:21:59PM Brad Bishop wrote:
I have a V10.1 system that usually takes 1-2 minutes to display the logon panel. This occurs in my application as well as signing into SYSPROG. There are no client databases attached in SYSPROG and there is no additional code installed in SYSPROG. Starting clients on the workstations and starting clients directly on the server behave exactly the same way.
When a client startup is delayed if a client startup is initiated the second session frequently presents the logon panel in seconds. Eventually the first session will present the logon screen.
If I look at the RevEngine the only three lines shown during the delay are:
RUN PS_INIT XXXXXX
RUN PS_SYSMON 'LOAD'
RUN RUN_EVENT ",'SYSTEM','SYSTEM','LOGIN*6*SYSPROG*LOGIN.SYSTEM*','1','SYSPROG'.SYSPROG',"
When I look at the Engine Info during the delay time Engine usage is always in the high 80's. When I look at the system performance the openinsight.exe process is consuming most of a singe processor.
Since this is happen without any application code running how do I track this down and get it resolved.
Thanks in advance.
bb
At 17 MAY 2022 03:47PM Andrew McAuley wrote:
Do you have an LH4.log hanging around? This is opened by the first person in who subsequently logs in slowly whilst additional login attempts cannot open it as it is exclusively opened by the first person so they proceed normally.
World leaders in all things RevSoft
At 17 MAY 2022 03:59PM bshumsky wrote:
Hi, Brad. We sometimes saw this problem when OI tried to clean up SYSAUTHLOG entries that are old, or exceeded the maximum number allowed. From the IDE TCL, can you COUNT SYSAUTHLOG and see if that table is really full? If it is, you might try to delete the entries from system monitor (you can't CLEARTABLE because it's a system table, but you should be able to DELETE_ROW SYSAUTHLOG,* ).
Please note that you should also rebuild the indices after deleting the rows - the fact that the indices didn't get properly updated when the table was cleaned up was actually the cause of the problem…
These issues have been fixed for 10.2…
Hope that helps,
- Bryan Shumsky
At 18 MAY 2022 11:13AM Brad Bishop wrote:
There were 1.2M records in SYSAUTHLOG. I cleared them and rebuilt the index. That appears to have resolved the issue.
The command to delete them from the system monitor was "RUN DELETE_ROW SYSAUTHLOG,*". Took a while and I had to rerun it a few times as it stopped part of the way through, but eventually it cleared all the records.
Thanks for the quick and accurate response.
At 18 MAY 2022 12:16PM bshumsky wrote:
There were 1.2M records in SYSAUTHLOG. I cleared them and rebuilt the index. That appears to have resolved the issue.
The command to delete them from the system monitor was "RUN DELETE_ROW SYSAUTHLOG,*". Took a while and I had to rerun it a few times as it stopped part of the way through, but eventually it cleared all the records.
Thanks for the quick and accurate response.
Great! Sorry for the trouble, but glad we were able to diagnose it.
FYI, as far as I know from OI 10.x system monitor, you don't need RUN before an stproc command - so it should work just as:
DELETE_ROW SYSAUTHLOG,*
- Bryan Shumsky