Last modified: 2014-06-07 02:57:44 UTC
The scap steps which run tasks on cluster hosts via ssh show a progress meter and display stack traces and other output from failed hosts, but often the hostname of the specific host where a failure is encountered is shown only in the middle of the output somewhere. This makes it harder than it needs to be for a deployer to determine which host is failing and report that failure in SAL or ask on irc if this is a known problem. Make the hostname more visible for each error, possibly by prepending the hostname to each line of output or at least the beginning of the error output. Also consider outputting a summary of failed hosts for a given task to the logs. This could even be automatically appended to SAL using the scap.cli.Application.announce() logging function.