xorbitsai / xorbits

Scalable Python DS & ML, in an API compatible & lightning fast way.
https://xorbits.readthedocs.io
Apache License 2.0
1.1k stars 67 forks source link

ENH: Add a status monitor to trace the running stage of subtask #677

Closed Flying-Tom closed 11 months ago

Flying-Tom commented 1 year ago

What do these changes do?

Add a status monitor to trace the running stage of subtask, which will be used to terminate stale tasks, e.g. when requesting mem quota

Related issue number

Fixes #489

Check code requirements

codecov[bot] commented 1 year ago

Codecov Report

Merging #677 (6919953) into main (95e964f) will decrease coverage by 35.10%. The diff coverage is 75.51%.

@@             Coverage Diff             @@
##             main     #677       +/-   ##
===========================================
- Coverage   93.51%   58.42%   -35.10%     
===========================================
  Files        1025     1025               
  Lines       79335    79379       +44     
  Branches    16442    16446        +4     
===========================================
- Hits        74187    46374    -27813     
- Misses       3462    30916    +27454     
- Partials     1686     2089      +403     
Flag Coverage Δ
unittests 58.32% <75.51%> (-35.09%) :arrow_down:

Flags with carried forward coverage won't be shown. Click here to find out more.

Files Changed Coverage Δ
...hon/xorbits/_mars/services/scheduling/api/oscar.py 72.58% <0.00%> (-27.42%) :arrow_down:
.../xorbits/_mars/services/scheduling/worker/quota.py 58.17% <0.00%> (-38.42%) :arrow_down:
...bits/_mars/services/scheduling/worker/execution.py 87.46% <75.75%> (-11.88%) :arrow_down:
...rbits/_mars/services/scheduling/worker/__init__.py 100.00% <100.00%> (ø)
...orbits/_mars/services/scheduling/worker/service.py 100.00% <100.00%> (ø)
python/xorbits/_mars/services/subtask/__init__.py 100.00% <100.00%> (ø)
python/xorbits/_mars/services/subtask/core.py 92.30% <100.00%> (-2.54%) :arrow_down:

... and 668 files with indirect coverage changes

aresnow1 commented 12 months ago

It's better to remove records if subtask is finished, they will take too much memory.

Flying-Tom commented 11 months ago

The title of pull request need to be changed.

Title has been changed