Issue
Standalone v1.93 system has about 1000 points being extended logged for site commissioning. There are "holes" in the extended log data.
Product Line
Andover Continuum
Environment
Standalone CyberStation v1.93
Cause
Standalone v1.93 system has about 1000 points being extended logged for site commissioning. There are "holes" in the extended log data.
The points are being logged at 2 minute intervals. Partner added a second (temporary) workstation to help with extended logging. This is still a standalone (SQLExpress) system. Extended log data is being extracted using Excel on the standalone server.
Even with the second CyberStation helping with the load, there are still holes in the log data.
Resolution
The workstations cannot handle the extended logging of ~1000 points at 2 minute intervals. Not using SQL Express on a single user even with a second workstation.
See also Extended Logging gaps, performance and load balancing.
This article mentions 2000 points per workstation but it is likely that this was all based on 15 minutes intervals and running full SQL Server.
At 2 minutes intervals you need to gather 720,000 entries/day or 8.333 entries/sec.
At 15 minute intervals you need to gather 96,000 entries/day or 1.111 entries/sec.
This is a big difference.
In addition, the Excel script is reading from the same extended log table being written to, at 8.333 entries/sec.
The controller log engine is completely independent of the extended log engine. The controller logs never send their data to CyberStation; CyberStation has to go get it. The extended log engine uses the general preference settings. In this case the settings are as follows and appear fine.
7. Extended log database check interval (minutes): 5
8. Extended log maximum LAN buffer interval (minutes): 30
10. Extended log automatic purge interval (days): 5
CyberStation is constantly trying to gather data from logs well before the “deadline” when the log fills and is about to be overwritten. While doing this, CyberStation also has to monitor time in the controllers, to make sure it is not too far out of sync, and this may potentially take place on field bus controllers on many network controllers, which may even be in different time zones. It also has to know when controllers have gone offline, or for some reason extended log values have not been updated.
In this case, CyberStation must try to play catch up when the controllers come back online (remember, these offline controllers have most likely continued to log data). And the same “catch up” has to take place if one or both workstations are turned off, or are busy doing other tasks like running reports or polling points via Pinpoint or Listviews.
The bottom line is that the CyberStation extended log data collection engine is sophisticated and complicated. To get a sense of this run AccTrace.exe (in the \Continuum folder) on a system busy with extended logs and you'll see it firsthand.
When a SQL Express system with one or two workstations is trying to collect the logs and do everything else that CyberStation does (listviews, reports, schedules, pinpoint, etc), at times, some data may get dropped. The recommendation at this site was to lighten up the CyberStation load by changing the log interval to 15 minutes and monitor if the extended log holes get eliminated.
