Conversation
common.sh produces *very* verbose debug output on failure. This leads to the actual error being masked as it is pushed out of the scrollback buffer by many pages of dmesg. This change redirects this output to a FAILURE log file which is timestapmed. Signed-off-by: Thilo Fromm <thilofromm@microsoft.com>
|
I like this a lot, but can we see this file in CI? |
CI could chose to |
Yeah, I've never found it useful. I don't want to see it in CI normally. If it's not exposed as an artifact, I wonder if there's any point having this data at all. |
|
I have this commit in my selinux-coverage branch, but I was wondering if this is the way to go. There, I'm pushing the build logs to bincache, but I'd need a reassurance that this is a safe thing to do. The commit in question is 8f02179. |
|
I think that the dmesg, df and mount were added by me because at some point we had a rather mysterious weird failures in CI, which I think just went away by themselves (probably got a Flatcar update or something). So maybe an option would be to just drop those. |
common.sh produces very verbose debug output on failure. This leads to the actual error being masked as it is pushed out of the scrollback buffer by many pages of dmesg.
This change redirects this output to a FAILURE log file which is timestapmed.