Open atdservicebot opened 4 years ago
From johnclary on 2019-05-10:
Response to TMC just now:
Thanks, all. Going to go ahead and include Cole and Chris in this discussion.
We pull the signal data from what's called the "STATUS" table in KITS. I believe there is another event-related table that we do not reference.
From the status table, we process any status change that is either a (1) conflict, (2) cabinet, or (3) comm outage.
We call cabinet flash "scheduled flash", because it's the status the signals send when they're on overnight flash. And it seems they send this status when the technician has manually put the signal on flash.
Cole and Chris—could you shed any light on what might explain discrepancies between the screen pictured in KITS vs our dashboard?
From johnclary on 2019-05-14:
From Cole R. on May 13:
--
We have gathered some notes related to these discrepancies.
Intersection Status Table
o Updated on 15 s intervals by the Status Service, only if a signal status changes
o As John mentioned, only Conflict, Cab Flash and Comm Fail are recorded in this table
o This table was integrated in older versions of KITS and is not used within the new interface and therefore has not been updated for alternate events
· Currently signal status in KITS is relayed through direct connections to the comm services
o Comm Failures override Conflict and Cab Flash in this table
· If a Conflict or Cab Flash occur, with a subsequent comm failure, the entry for a given signal will show comm fail
· Alternately, KITS will maintain the alert for Conflict while also listing the Comm Fail separately
o Slight correction to email, Status values within table are: Cab Flash = 1 and Conflict = 2
· Stop Time
o Current issue in which Stop Time is reported when a signal is in Flash
· Joshil and I have been unable to reproduce in test environments
· In nearly every example onsite, comm failures are prevalent during or near the Stop Time
§ A Conflict is recorded when both the Stop Time and Cab Flash bits are active in the poll message that is received from the controller
o Not recorded within Intersection Status table, therefore not included in Data Tracker counts
I looked up the event log in this timeframe for a few of the signals and attached them. It appears the Traffic Signal Monitor page is updating every 2 minutes after keeping an eye on it today. I was hoping to find some live discrepancies to review but so far they have been in sync, see attached screenshot. It appears the discrepancies seen during the storm could be a combination of issues as mentioned, Stop Time being reported though not included in Status table, Comm failure overrides, and potential delay during the 2 min interval and updating cycle from Status service. I will continue to monitor.
From johnclary on 2019-05-14:
Based on this info: if we want the signal monitor dashboard to more closely mirror what's shown in KITS we need to update the app to use the KITS event log instead of the status table
From dianamartin on 2019-06-06:
updated communication from Jim to Ben. Ben's response:
We spoked just a moment ago about the Traffic Signal Monitor map missing some signals that were in flash. I spoke with Diana and she let me know that this is a known issue and is on their list of projects to take on in the near future. She also let me know that this is due to this map pinging the incorrect KITS status server.
From Lance B:
Jen discovered a discrepancy between the flash status on the Traffic Signal Monitor page (http://transportation.austintexas.io/signals-on-flash/#) and our actual flash status in KITS. This became very evident during the storm when we had only 1 Unscheduled Flash shown on the Traffic Signal Monitor page and we had over a dozen signals in flash (“Conflict” status in KITS) for over a period of time ranging from half an hour to almost two hours.
Please let us know how we can assist with any information you might need. Thanks!
Migrated from atd-data-and-performance #270