Conversation
Pull Request Test Coverage Report for Build 22773699455Warning: This coverage report may be inaccurate.This pull request's base commit is no longer the HEAD commit of its target branch. This means it includes changes from outside the original pull request, including, potentially, unrelated coverage changes.
Details
💛 - Coveralls |
d5bdeaf to
f2e27f5
Compare
|
@pauldg is also interested in this |
|
Hi @nuwang , this is going to be very useful!
For |
|
|
Line 25: entity combining - the combined entity has Lines 52-68: Maybe the abstract destinations could be left out. There is a lot of other good info in here and the fact that entities do not match with abstract destinations is not interesting. Line 70: slurm-training is rejected for a tag mismatch but it’s not clear why. It would be better if all tag categories (accept/prefer/require/reject) were listed for the entity, and if tags were separated into categories for the destination. The reason that there is a tag mismatch is that slurm-training requires the ’training’ tag and the entity does not have the ‘training' tag, but it is not obvious from this explanation. Lines 137-141: There is something odd here because pulsar-mel3 is listed twice and pulsar-QLD also matched. The ranking function being used by the job conf in this case is weighted_random_sampling. Everything else looks fantastic! |
|
Thanks @cat-bro, that was super useful feedback. I think all the issues you highlighted have been addressed now. The last one was particularly interesting - because it's a consequence of using weighted random sampling without weights being defined. As a result, the same destination is considered again when making the next random choice. I've changed it so that, if weights are not defined, it falls backs to standard random sampling (without replacement). |
This PR adds support for
closes: #153