|
|
During CI, it has been observed that some processes didn't leave their
wait state. (So they were "hanging")
During analysis of the event log on that CI node, we found out that at the
same time those processes entered this state, the NTP client set a new time
and indeed the local time of the CI node changed (it was
changed 2 hours forward in our case).
For the record, the processes we observed this with was jom.exe and
cc1plus.exe (the latter is the g++ compiler in the mingw toolchain)).
We therefore have a strong suspicion that the NTP time change was causing
the CI hang, so we simply disable it.
The virtual machine should anyway get the correct time from its VM host
when its started.
Change-Id: I1204063399e2a4741b7b26b79858759c1bc5eab2
Reviewed-by: Jędrzej Nowacki <jedrzej.nowacki@qt.io>
|
