Also, added slightly better error output for certain types
of test failures. Finally updated splunkbase app defaults in
validate_args, but they are commented out for now while we
use the S3 versions.
in the output files when a search generates certain types of errors.
For example, an error where it cannot reach the Splunk endpoint
server. These are rare, but good to have.
GH Action Machine config with the
mock option. Also, shuffling detections after they are put into a
list to distribute runtime and load as
much as possible.
we know that we will not have a test for it. For example,
if we try to start 4 containers by have only 2 tests at
the beginning we will only start 2 containers. This saves
a lot of startup time and resources. There is a descriptive
printout for this as well.
Also, bumped the maximum startup time for 6 minutes to
10 minutes. This, combined with the system info from
the previous commit should let us determine if we can
bump the number of containers per GitHub Actions
VM from 1 to 2 or more.
This will help users dianose if their systems are overburdend
and also helps us figure out what the appropriate number
of containers to run on cloud infrastructure, like GitHub
Actions, may be without logging directly into the machine
doing the testing. In some cases, we can't log into those
machines by design. If you're running tests at home, you
can also just listen to the volume of your computer's fans.
apps stored on S3. Pointed at the PATCHED linux_sysmon.
Added a slight delay in between starts of containers for
performance reasons. Enabled better handling and ability
to download http_path local apps to a folder instead of
passing them in as string for the container to download.
everything smoother. The largest differences
are improving detection of containers that
take too long to start (or that crash while)
they are starting by adding a timeout. That
timeout is set to 360 seconds and might need
to be tuned in the future to a larger number
since this was tested on a fast machine with
fast network. The other large change is an
initial pass at updating and committing
detections that have passed the test back
to the repo. This still needs a lot of testing and
refinement.