TC_PCC_2_1 fails on darwin

[andy31415]

Problem

Seems like a timeout waiting for log content mDNS service published.

Multiple occurences in PRs like:

• https://github.com/project-chip/connectedhomeip/runs/8046892035?check_suite_focus=true
• https://github.com/project-chip/connectedhomeip/runs/8063594126?check_suite_focus=true
• https://github.com/project-chip/connectedhomeip/runs/8071586794?check_suite_focus=true

Or even ToT (less frequently it seems):

https://github.com/project-chip/connectedhomeip/runs/8047808378?check_suite_focus=true

Actual reproduction steps are unclear. @andy31415 flags seem to be:

export TSAN_OPTIONS=halt_on_error=1
export LSAN_OPTIONS='detect_leaks=1 suppressions=scripts/tests/chiptest/lsan-mac-suppressions.txt'
export MallocNanoZone=0

to have similar logging like CI, however could not reproduce. It may be dependent on multiple runs or some other CI-specific setting.
Logs generally seem to contain some 10s gaps:

...
02:14:09.248 - APP  OUT  : [1661566449248] [37954:150690] CHIP: [ZCL] PCC Server: Constant Temperature Supported
02:14:23.030 - APP  OUT  : [1661566449248] [37954:150690] CHIP: [ZCL] PCC Server: Constant Speed Supported

and

14:32:05.916 - APP  OUT  : [1661783525916] [80180:216222] CHIP: [ZCL] Initialized new lock door endpoint ...
14:32:22.576 - APP  OUT  : [1661783525916] [80180:216222] CHIP: [ZCL] Window Covering Cluster init

which seems to indicate maybe a problem with the python code receiving buffered output or not receiving output until some type of wake up event (logs are not truncated as far as I could tell).

Proposed Solution

Figure out the reason and fix.

[andy31415]

If we get a consistent reproduction PR, we may want to try to use stdbuf -o0 or similar in scripts/tests/chiptest/runner.py to see if that improves things.

[andy31415]

Closing for now: i Have not seen TC_PCC_2_1 fail in a while now. Maybe the stdout changes took effect.