HW CI is not reliable enough

[rusefillc]

Humans are fucking expensive, we commit a lot. 20 seconds waste of flaky test multiplies.

[rusefillc]

for example https://github.com/rusefi/rusefi/actions/runs/3353650060/jobs/5556605696#step:16:1340

[rusefillc]

@mck1117 any chance you could help?

https://github.com/rusefi/rusefi/actions/runs/3451774157/jobs/5761229627#step:16:1154

[rusefillc]

https://github.com/rusefi/rusefi/actions/runs/3481584396/jobs/5822865985#step:16:1151 @mck1117

[rusefillc]

@nmschulte this one is a functional defect

not sure if proteus HW_CI setup has external wires like discovery setup has

https://github.com/rusefi/rusefi/actions/runs/3505546092/jobs/5871917337

[mck1117]

They both have external wires. The comments in the relevant Java files tell what mods are needed.

[mck1117]

Those fails all look like the same thing - comms time out, which may mean that the ECU is locked up. What's the failure rate like?

[rusefillc]

i want to say 5%-ish?

[rusefillc]

@mck1117 still 5% failure rate

https://github.com/rusefi/rusefi/actions/runs/3858743431/jobs/6577597618

[rusefillc]

@mck1117 https://github.com/rusefi/rusefi/actions/runs/3888361986/jobs/6635598767

[rusefillc]

@mck1117 https://github.com/rusefi/rusefi/actions/runs/3889093346/jobs/6637049388

[rusefillc]

@mck1117 https://github.com/rusefi/rusefi/actions/runs/4036174871/jobs/6938589005#step:18:726

[mck1117]

if we increase that 500ms timeout, does it get better?

[rusefillc]

@mck1117 what do we increase to 500ms exactly? Everything seems to be at least 500ms already?

[mck1117]

[rusefillc]

Sorry did not read it right

[mck1117]

but it is strange because just above that it says 5000ms

[rusefillc]

@mck1117 i believe different method invocations come with different timeouts