rpmbuild tracking resource/quality metrics
rpmbuild should start tracking cpu/memory/disk usage for
use by build factory schedulers.
Some slight plumbing changes (i.e capturing gcc stderr
rpmbuild -ba foo.spec 2>build.log
provides raw data on the number of warnings/errors seen during
a build. Its feasible to include warning/error counts in binary packages,
as well as the error (or even the full) log in SRPMS's, for proper RPM_OPTFLAGS choice.
Its also possible (but trickier because convolved with package generation/signing) to
run rpmlint-like on either the build-root, or on the accumulated metadata, and include
the warnings in binary/source packaging.
Blueprint information
- Status:
- Started
- Approver:
- Jeff Johnson
- Priority:
- Medium
- Drafter:
- None
- Direction:
- Approved
- Assignee:
- Jeff Johnson
- Definition:
- Discussion
- Series goal:
- Accepted for 5.3
- Implementation:
-
Slow progress
- Milestone target:
- None
- Started by
- Jeff Johnson
- Completed by
Related branches
Related bugs
Sprints
Whiteboard
Some of the build metrics:
1) status of build scriptlets
2) build scriptlet timers
have been prototyped. It's unclear whether the
metrics should be in packaging (or not).
A mongodb store is a perfectly acceptable alternative/
the issue is that anyone who chooses _NOT_ to use mongo will _NOT_ benefit
if the info is _NOT_ in *.rpm packages somehow.
(aside)
Needs a clearer more positive statement of the design issue ...
