I am running a 2 server redundant system with PMI, FSQL, and Reporting plug in installed on both. Yesterday in FSQL I ran the pre-installed Clean Alarmlog group. On a simular note, i also kicked over to the second server in FSQL (redundancy is configured between the two).
After these two events occured, it seemed as if I just totally stopped logging any alarms. I went into FSQL and looked at my alert settings on both machines. Neither one has alarmlog nor alarmstatus typed in for the tables, which i found to be very odd. I typed in the table names on both machines.
This morning when we looked at the alarm history, the same alarm was on the list multiple times. What would cause this???
I preformed a truncate statement on the alarmlog which effectively purged it. After I purged it, I still had about a dozen alert bits that should have generated alarms. Will I only get an alarm on a transition from false to true (assuming my bits all alarm on high state)???
We have had two new alarms since and they are working as expected. Any other insight anyone can offer into how the alarming mechanism works???
Thanks
Hi,
A few elements of what you described struck me as strange. First off, the missing table names. What is the version of FactorySQL on your machines? Those settings are set locally and not shared through redundancy, so I’m not sure how they’d get wiped out.
Next, it’s important to note that the “clear alarm log” group you mentioned is not pre-installed in the sense that it’s meant to be used in general, it just happens to be a group included with the demo project. It might work for you, if you happen to have the alarm log and status tables named as it would expect, but you’ll want to look closely at it and make sure it’s doing the correct thing in your case.
For entries in the log- they are only entered in the log when the event occurs. So, if you delete everything, you won’t get an entry until the alert goes active again. You might get an entry when the alert clears, if it’s starting from an active state, if the alert is set to report clear messages.
As for why you got multiple entries- are there any messages in the FactorySQL log (Help->Log Viewer) that might indicate that the server/backup switched? Perhaps they went back and forth before settling down, and each registered an alert (which would happen when the group started logging). I would recommend upping your node timeout to at least 15 sec, as I think the default 5 seconds is too short.
Hope this helps a bit, please post back if more weird stuff occurs.
Regards,
I was able to modify the alarms table sql query to only show one instance of the alarm. I did some looking around in FSQL in the Log Viewer. Just this week, I have 42 events for Project State Changed: Project state has been changed to active and 42 events for Responsibility state changed: Responsibility has changed to Master_Negotiated. My best guess for the cause of this is in the Redundancy Settings and OPC Quality Monitoring. I am using OPC Quality to determine node state. My OPC server is KEPServerEX V5.
Is it reasonable to think that this is the cause for the project state and responsibility changing so often. Perhaps I need to look at my scan settings in kepware? What are your thoughts?
Thanks