scheduler: do not balance the empty regions #3344
Conversation
Codecov Report
@@ Coverage Diff @@
## master #3344 +/- ##
==========================================
- Coverage 74.94% 74.91% -0.04%
==========================================
Files 243 243
Lines 23245 23250 +5
==========================================
- Hits 17421 17417 -4
- Misses 4259 4265 +6
- Partials 1565 1568 +3
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
|
rleungx
added
do-not-merge/work-in-progress
server/schedulers/balance_region.go
Outdated
| @@ -167,6 +171,13 @@ func (s *balanceRegionScheduler) Schedule(cluster opt.Cluster) []*operator.Opera | |||
| } | |||
| log.Debug("select region", zap.String("scheduler", s.GetName()), zap.Uint64("region-id", region.GetID())) | |||
|
|
|||
| // Skip the empty region | |||
| if region.GetApproximateSize() <= core.EmptyRegionApproximateSize && regionCount > balanceEmptyRegionThreshold { | |||
There was a problem hiding this comment.
it seems better to define a region filter and filter empty regions in Rand{Pending, Follower, Leader, Learner}Region.
Signed-off-by: Ryan Leung <rleungx@gmail.com>
Signed-off-by: rleungx <rleungx@gmail.com>
rleungx
force-pushed
the
unbalance-empty
branch
from
17aeec9 to
718b0a3
Compare
Signed-off-by: Ryan Leung <rleungx@gmail.com>
rleungx
force-pushed
the
unbalance-empty
branch
from
718b0a3 to
8a44024
Compare
ti-chi-bot
added
status/LGT2
Signed-off-by: Ryan Leung <rleungx@gmail.com>
rleungx
force-pushed
the
unbalance-empty
branch
from
e513b6f to
5219763
Compare
|
/merge |
|
@disksing: It seems you want to merge this PR, I will help you trigger all the tests: /run-all-tests Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the ti-community-infra/tichi repository. |
|
This pull request has been accepted and is ready to merge. Commit hash: 5219763
|
ti-chi-bot
added
the
status/can-merge
|
@rleungx: Your PR has out-of-dated, I have automatically updated it for you. At the same time I will also trigger all tests for you: /run-all-tests Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the ti-community-infra/tichi repository. |
What problem does this PR solve?
After
drop tableor such kind of operation, there might be a large number of empty regions which cannot be merged immediately. The balance region scheduler will balance them if the stores' size have the difference. In this case, it may cause a lot of balance operation which affect the performance.What is changed and how it works?
This PR will skip the empty region when selecting the region.
Check List
Tests
Release note