Increase test timeouts for slow CI/build virtualization
okeuday opened this issue · comments
10 seconds is too short for some of the virtualized CI/build instances that run CloudI tests. Will need to be increased to 30 seconds or higher. 60 seconds may be best.
cat /sys/devices/system/clocksource/clocksource0/current_clocksource
cat /sys/devices/system/clocksource/clocksource0/available_clocksource
Will show kvm-clock
for KVM pvclock (xen
on AWS instance, with tsc
,hpet
,acpi_pm
available, the typical options with a Linux installation, AWS EC2 VMs that use Nitro should have kvm-clock
as the clocksource). May be good to get that data into code_status
through some route.
erlang:system_info(os_monotonic_time_source).
erlang:system_info(os_system_time_source).
Currently provide the C call info. LXC container is currently the problem.
virt-what may be used to determine the virtualization and disable tests that require reasonable clock_gettime
CLOCK_MONOTONIC
return values, though it would need to be a build/test time dependency due to the GPL use:
git://git.annexia.org/virt-what.git
https://people.redhat.com/~rjones/virt-what/
Ok, with the --with-test-timeout
and --with-long-test-timeout
configure script arguments this problem is fixed.
For reference: https://gitlab.alpinelinux.org/alpine/aports/-/issues/11474