kfaraz opened a new pull request, #14584:
URL: https://github.com/apache/druid/pull/14584
### Description
With the recent improvements to `CostBalancerStrategy`, we can now move more
segments in each run.
### Changes
- No change if `smartSegmentLoading` is disabled
- If `smartSegmentLoading` is enabled, the following computations take place.
- Based on the number of used segments, calculate the number of
`balancerComputeThreads` as explained below.
- Based on usage skew and total segments on all active historicals in a
tier, calculate the `maxSegmentsToMove` as follows.
- Emit some additional metrics and logs
### Calculations
#### `balancerComputeThreads`
As established in #14484 , 4 balancer compute threads were able to perform
cost computations for 1000 moving segments in a cluster of 1 million total
segments in about 5s. Thus,
```
total computations with 4 threads in 5s = 1k * 1M = 1B
total computations per thread in 5s = 250M
add a buffer of 5s
total computations per thread in 10s = 250M
total computations per thread in 80s = 2B
maxComputationsPerThread (per coordinator cycle) = ~2B
```
#### `maxSegmentsToMove` per tier
```
totalSegments = all segments that would participate in cost computations
totalSegments = all queued and loaded segments on all active historicals
totalComputations = maxSegmentsToMove * totalSegments = numThreads *
maxComputationsPerThread
maxSegmentsToMove = (numThreads * maxComputationsPerThread) / totalSegments
```
#### `segmentsToMove` in a given tier
```
maxUsagePercent = highest (projected) disk usage of an active historical
minUsagePercent = highest (projected) disk usage of an active historical
usageFactor = Math.max(10, maxUsagePercent - minUsagePercent)
segmentsToMove = maxSegmentsToMove * usageFactor
```
### Classes to review
* `SegmentLoadingConfig`
* `BalanceSegments`
* `TierSegmentBalancer`
### Release note
> TODO
<hr>
<!-- Check the items by putting "x" in the brackets for the done things. Not
all of these items apply to every PR. Remove the items which are not done or
not relevant to the PR. None of the items from the checklist below are strictly
necessary, but it would be very helpful if you at least self-review the PR. -->
This PR has:
- [ ] been self-reviewed.
- [ ] using the [concurrency
checklist](https://github.com/apache/druid/blob/master/dev/code-review/concurrency.md)
(Remove this item if the PR doesn't have any relation to concurrency.)
- [ ] added documentation for new or modified features or behaviors.
- [ ] a release note entry in the PR description.
- [ ] added Javadocs for most classes and all non-trivial methods. Linked
related entities via Javadoc links.
- [ ] added or updated version, license, or notice information in
[licenses.yaml](https://github.com/apache/druid/blob/master/dev/license.md)
- [ ] added comments explaining the "why" and the intent of the code
wherever would not be obvious for an unfamiliar reader.
- [ ] added unit tests or modified existing tests to cover new code paths,
ensuring the threshold for [code
coverage](https://github.com/apache/druid/blob/master/dev/code-review/code-coverage.md)
is met.
- [ ] added integration tests.
- [ ] been tested in a test Druid cluster.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
