unattended_install: Two tweaks to attempt_to_log_useful_files #2489
Conversation
Currently when the command is not available on the target distro another loop will be initiated on the next available session (serial console). Let's just rely on the files to prove the session is stable and ignore the status for commands. Signed-off-by: Lukáš Doktor <[email protected]>
This dump is very useful when looking for udev issues. Signed-off-by: Lukáš Doktor <[email protected]>
| @@ -1217,7 +1217,7 @@ def attempt_to_log_useful_files(test, vm): | |||
| fd_dst.write("Unknown exception while getting " | |||
| "content: %s" % details) | |||
| failures = True | |||
| for cmd in ["journalctl --no-pager"]: | |||
| for cmd in ["journalctl --no-pager", "udevadm info --export-db"]: | |||
There was a problem hiding this comment.
@ldoktor cmd run once is enough, That code should not be in the outermost loop
There was a problem hiding this comment.
In case there are no failures, the commands will only be executed just once. Otherwise it makes sense to execute them multiple times as it's likely they failed to execute as well (this usually happens when net becomes unusable).
There was a problem hiding this comment.
This may be a question I've asked before (during the initial feature review), but doesn't this make it possible to overwrite dst file in a different iteration of the loop (over sessions)? Shouldn't this:
- Check if
dstexists - If not, run the command
- If successful, create the
dstfile with the content from the command
There was a problem hiding this comment.
No, each iteration uses a different direcotry (see line 1221). In case of a failure it might be useful to have multiple copies and usually it PASSes so one ends up with a sigle dir.
The commands might not be available on the systems or might fail for a different reason so let's just rely on errors from files logging. Also add "udevadm" dump on failure.