Closed datid closed 7 years ago
So, the more interesting thing would be to see the restart after this. The 1556 error indicates that we have closed something before we intended to. Not certain that this is exactly related to the Meerstetter, but it gives us a starting point. The best way to approach this is to stop and start the program until we get this issue. Can we leave the Meerstetters on possibly so I can look at this?
Closed lo-co/exscalabar#129 that indicated that the UI was the issue.
This appears to be caused by the serial session closing. Not sure what is going on, but if the serial connection does not close down properly and you just restart, then you can expect things to hang. The question is why the TECs are not kicking out at start to prevent the system from not coming up properly.
Adding message to Serial Session::clear Session Data
[serial] Closing port associated with session {session_name}.
The error is not fed into this one so that we will know if an error is thrown when an attempt to close a session is made.
Added message in Serial Session::Configure Session
indicating which port we are kicking off a session for. Message is
[serial] Opening serial port for session {session_name}.
We don't want to handle the serial error here if there is one as this will cause the system to kick off the actor.
Something looks backwards in the logs here - I get a device stopping message which would be generated by the actor and then I get messages associated with setting up the TECs for shutdown. I will investigate further tomorrow.
Regardless, it looks like the session queue is killed prior to the final TEC shutdown code being called.
Able to reproduce behavior by starting the program and then cycling the instrument on and off via the UI several times (3). Each time, I attempted to save a file. Third or fourth time it failed.
So, this is not clear whether this is a device registration issue or whether this is an actual file saving issue. When the save button is hit, the user should get a message that looks like
[file] New file opened at u:\exscalabar\Data\test_20170630_200908.txt.
When it doesn't start saving, you don't get this message.
The message
[ctl] Save state set to {boolean}.
is generated but the Save Data MSG::Do
in the Controller
library and is sent by the file save web service VI. What this will do is (1) toggle the save state in the CVT and (2) send a message to both the main File Actor
and the mirrored FileActor
if there is one.
Sooo...totally broke this with a restart. Now it doesn't even give the message.
Cycling stop-start got the save state back and a file writing (although no mirror).
Added a Sys Log hook in the serial session. It is opened in the configuration routine and closed in the clear Session Data
routine. Now there are some new messages that have the tag [serial]
.
Still getting errors at the closing of each of the serial sessions. Not sure what is going on here, but if they are not closing properly then this could be a big issue.
Appears to be an issue with both the MFCs and the Meerstetters shutdown sequence. The Serial Session
property is closed with the first call to close the session.
Another issue that I found was that the Vaisala code never called a close on the serial port. This may have been the real issue, but I can not confirm this right now. I have changed the method Vaisala::Close
to the override method Vaisala::Stop Device
.
I think the above fix worked. Stopped and started 10x with no fails - wrote to file every time. Will be closing this issue with next check in. Still having an unrelated issue with the MFCs and Meerstetters.
Though the save mode is set to True it sometimes fails to save data.
Could be a registration problem, possibly caused by a shutdown issue
Not all the meerstetters seem to stop properly -