Merge lp:~kissiel/checkbox/speed-optimization into lp:checkbox
Proposed by
Maciej Kisielewski
Status: | Merged |
---|---|
Approved by: | Maciej Kisielewski |
Approved revision: | 4520 |
Merged at revision: | 4517 |
Proposed branch: | lp:~kissiel/checkbox/speed-optimization |
Merge into: | lp:checkbox |
Diff against target: |
71 lines (+6/-8) 4 files modified
checkbox-ng/checkbox_ng/misc.py (+1/-1) plainbox/plainbox/impl/session/state.py (+1/-1) plainbox/plainbox/impl/unit/test_unit.py (+0/-3) plainbox/plainbox/impl/unit/unit.py (+4/-3) |
To merge this branch: | bzr merge lp:~kissiel/checkbox/speed-optimization |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Po-Hsu Lin | Approve | ||
Paul Larson | Approve | ||
Review via email: mp+307553@code.launchpad.net |
Description of the change
This MR is the first (and simple) patch set that improves plainbox startup time.
The speed-up depends on how many jobs there are in your provider 'database'.
Here are benchmarks from my slow NUC:
1000 jobs - 44s -> 24s
2500 jobs - 2m43s -> 1m19s
The benchmarks were done using this tool:
https:/
The tool generates providers and checkbox-cli launcher that doesn't run any of the jobs. Measuring wall time of running this launcher gives an estimate how long checkbox takes to 'spool up'.
To post a comment you must log in.
Wow, part of me really wants to pick this apart and find if there are even more gains, compare it to normal sets, and things like that, but I see no reason to block such an obvious improvement for any time at all. It's hard to argue with those results, this is great!