search

PX4 / PX4-Autopilot

PX4 Autopilot Software
https://px4.io
BSD 3-Clause "New" or "Revised" License
8.57k stars 13.55k forks source link

[TEST] Disable Multi-EKF for SITL to see if CI failures are related #24017

Open MaEtUgR opened 4 days ago

MaEtUgR commented 4 days ago

Solved Problem

https://github.com/PX4/PX4-Autopilot/issues/23965#issuecomment-2491827454 I think multi EKF is at least related since missing data is only possible if the selector message is not available or the status of the primary estimator: https://github.com/PX4/PX4-Autopilot/blob/2f65644aeb5cb4c2023e43ac45d8a9b01d0584ba/src/modules/commander/HealthAndArmingChecks/checks/estimatorCheck.cpp#L74-L100 and I checked a failing log, both topics are there. So where does it fail? Too many topics? There's only one instance of EKF in the log.

grafik

Now that I have a closer look the error is probably gone without multi EKF because it's always there before the selector data is available 🤦‍♂️ Why is the selector data not available earlier?

Test coverage

Let's see 👀

MaEtUgR commented 4 days ago

All SITL tests first run passed, re-run again (Collision Prevention pr also passed once). Some MAVROS test is not happy, I'll check that later.

MaEtUgR commented 4 days ago

At least the airspeed issue is not related to multi-EKF: grafik https://github.com/PX4/PX4-Autopilot/actions/runs/11959255538/job/33341396494?pr=24017 https://logs.px4.io/plot_app?log=e1cd1491-1093-495b-9e08-aa921b5063ef Let's do some more runs to see if the other issue is.

dagar commented 3 days ago

Let me see if we can improve/fix the "ekf2 missing data" startup race condition.

mrpollo commented 3 days ago

SITL tests seem to be passsing https://github.com/PX4/PX4-Autopilot/actions/runs/11975522906?pr=24017