search

Team2168 / 2015_Main_Robot

Source code for the 2015 season
2 stars 0 forks source link

Determine why communications are lost #69

Closed jcorcoran closed 9 years ago

jcorcoran commented 9 years ago

Periodically the robot loses communications with the driver station, after about 10minutes of running. This happened numerous times on Monday running master in the practice chassis.

Without changing master code, attempt to replicate the problem and observe comm loss with driver station. Verify you can ssh into the roborio when this condition is active.

Ssh into the roborio and monitor memory usage over time with the robot enabled. If it climbs without dropping we have a memory leak. Record some metrics on memory usage/time.

Use a laptop that has the latest version if the DS installed 08022015 I believe is the latest.

Debug test: Try disabling the console printer entirely and running the robot enabled without the smart dashboard open on the driver station. Attempt to replicate the problem.

NotInControl commented 9 years ago

So we noticed this a lot yesterday as well. I don't remember us having these issues before, or at least it wasn't this significant.

I briefly looked at the DS logs yesterday, we are running at about 75% cpu, and out trips times are good, but we have huge amounts of packet loss from time to time.

We also have issues with smart dashboard, I think that is a separate issue, but I couldn't get it up. I think part of the problem is that we have multiple computers connected to the Robotbot, so all of our tests should be done with only one computer on the network, and then add the others in to make sure they don't break it.

The comm losses were for short durations of time, 1 sec here and there, while I was ssh'ing into the RoboRio last night, I can not confirm or deny the ability to ssh into the RoboRio while a comm loss was experienced. A easier thing to do might be to do a running ping from the DS to RoboRio and watch that throughout the night.

I will make a point to do more testing around this in the next 2 days

jcorcoran commented 9 years ago

While I have seen intermittent comms loss, that resolves itself in a few seconds, I've also seen loss of comms that never comes back. I had to restart the rio.

I've observed both as recently as Monday. I had only one laptop connected to the robot. I was using an older version of the DS.

This latter failure seems like a problem in the user app to me. Hence the suggestions to look into SD and file I/O. I noticed the file I/O for Victors logger doesn't ever call flush, not sure if that's a problem or not.

jcorcoran commented 9 years ago

Kevin reported issues going away after having replaced the bridge with one from the KOP. Can't wait to have a more reliable piece of hardware replace this radio...

Closing this issue until we see it again.