DO 4 cycling On-Off
The Digital Output 4 is a visual indicator that GRIDlink is not only connecting to the Demand Response Server but is also parsing payloads properly. If there was a configuration or software problem DO4 would not be on at all. There are a number of things that can cause this to happen:
1. The DR Server is not responding to every request from GRIDlink.
This could be due to excessive load or there may be maintenance being performed.
This is a 24 hour trend of the Quality of Service (in blue) and parses (in red). Parsed are binary represented by either 1 (successful) or 0 (fail). Here there were 9 individual fails through out the day with a cluster of 5 at 22:00 in the evening. By simply reviewing the logs of all GRIDlinks in this service area and seeing the same cluster it is clear that there is maintenance being performed each night at 22:00. The other 9 fails are random. This is to be expected among Internet communications and since there are approximately 1440 parses each day, this represents a failure of ~0.6% and would not affect participation in any Event.
This could be as simple as a bad Ethernet cable to the network is experiencing excessive loads. Check with IT to see if the company using larger than normal amount of bandwidth as seen when performing large data dumps to corporate HQ.
3. Cell Connection
If connecting with a cell modem, it is not uncommon that the local cell tower is experiencing excessive loads, undergoing maintenance which is temporarily reducing bandwidth or even interference.
4. Wireless Network
If the local network connection is reliant on Wi-Fi or Spread Spectrum radios, it is very common to see miss parses due to latency and retries.
Load Averages
One way to detect a network problem is to see the load average on the Status page in GRIDview.
Normal
This is what the load should be:
High
Here is an example of a network issue which requires more than the usual number of retries. This increases the processor load.
This load will increase the number of failed parses by not enough to miss an Event. When it exceeds 4.5 for long periods, then the problem should be addressed.