"Add new jobs (Search)" should be able to filter by tier
Categories
(Tree Management :: Treeherder, enhancement)
Tracking
(Not tracked)
People
(Reporter: dbaron, Unassigned)
Details
When adding new jobs in treeherder (or, for that matter, using the same UI in mach try chooser), I'd like to be able to filter jobs by tier. In particular, what I'd often like to do is run all or many of the Linux64 tier1 test jobs (since Linux64 tests are run in AWS and have much more availability). But because of that availability, there are lots of extra test jobs for linux64 (e.g., fission variants, serviceworker redesign variants) that don't exist for other platforms and that are currently tier2. It's currently a bit of a pain to pick out "all the tier1 tests". I think some people just run 4x the number of tests relative to what they really want, because it's easier -- but it would be nice to be able to filter by tier in "Add new jobs (Search)".
(I'd also note that the way the searching itself works isn't entirely clear to me; it might also be nice to explicitly distinguish between "and" and "or" searches rather than doing what it seems to be doing which is both, but prioritizing the "and" results at the top.)
Comment 1•6 years ago
•
|
||
The runnable jobs API that we're querying to get the list of available jobs doesn't include the tier information. We'd have to retrieve the tier information from some other source and then combine them into one list to filter out non-tier1 tasks.
(I'd also note that the way the searching itself works isn't entirely clear to me; it might also be nice to explicitly distinguish between "and" and "or" searches rather than doing what it seems to be doing which is both, but prioritizing the "and" results at the top.)
Weird, with the options set up the way they are, tokenize + matchalltokens together should require the results to match all of the terms being searched (so it should be an AND search, not an OR search). Maybe the threshold needs to be lowered from the current 0.6 down to something like 0.5 or 0.45 and see if that gets rid of some of the extra matches.
It's also supposed to be sorting the results alphanumerically every time it searches, but that clearly isn't always happening. Not sure why.
There were a lot of additions to the job finder I wanted to add but never had time to work on (more explicit search operators like try fuzzy uses, etc). I'm busy doing Iris work these days and won't have a chance to look at this anytime soon. Someone else can take this. All of the available options for the fuzzy search package I'm using are described here, it should be pretty easy to play around with it and get it to do what you want.
(If the runnable jobs API was changed to include the tiers for each job, it would be much easier to add a filter for them in the fuzzy finder...)
Comment 3•6 years ago
|
||
They would need to be added to the file located here (as one example job):
https://taskcluster-artifacts.net/Jj3mjfXkSAihTXJXflqhyg/0/public/runnable-jobs.json
Dustin-- Is it possible to add tiers to this file?
Comment 4•6 years ago
|
||
Should be -- a bug in firefox build system :: task graph is the place to start.
| Assignee | ||
Updated•4 years ago
|
Description
•