[fix #623] region may be missed

[TrafalgarRicardoLu]

What problem does this PR solve?

Issue Number: close #623

Problem Description: If there is a region split between two queries, it may miss some regions.

What is changed and how does it work?

I change the time of getting region from before scan to after scan.

When we build client, we will use region info in regionCache. After region split, it's outdated.
But when client try to scan the region, it will get the newest region info from pb.

In the original code, we get region info before scan. So the region info is still outdated and it will use the old endKey to get next region. Using the old endKey may miss some regions that split during the two query.

I get the region info after scan to make sure the return value is newest.

Check List for Tests

This PR has been tested by at least one of the following methods:

• Unit test
• Integration test
• Manual test (add detailed scripts or steps below)

[shiyuhang0]

@iosmanthus PTAL

[codecov]

Codecov Report

Merging #624 (6da3e99) into master (3724e87) will decrease coverage by 0.02%.
The diff coverage is 75.00%.

@@             Coverage Diff              @@
##             master     #624      +/-   ##
============================================
- Coverage     34.76%   34.73%   -0.03%     
- Complexity     1421     1424       +3     
============================================
  Files           278      278              
  Lines         17356    17358       +2     
  Branches       1972     1972              
============================================
- Hits           6034     6030       -4     
- Misses        10713    10718       +5     
- Partials        609      610       +1     

Impacted Files	Coverage Δ
...ommon/operation/iterator/ConcreteScanIterator.java	0.00% <0.00%> (ø)
...g/tikv/common/operation/iterator/ScanIterator.java	78.94% <ø> (ø)
...ikv/common/operation/iterator/RawScanIterator.java	83.33% <100.00%> (+0.57%)	⬆️
...java/org/tikv/common/region/RegionStoreClient.java	52.49% <100.00%> (+0.06%)	⬆️
...rc/main/java/io/grpc/netty/NettyClientHandler.java	56.89% <0.00%> (-5.61%)	⬇️
...ty/handler/codec/http2/Http2ConnectionHandler.java	49.63% <0.00%> (+0.97%)	⬆️
.../main/java/org/tikv/common/region/RegionCache.java	50.92% <0.00%> (+1.85%)	⬆️
src/main/java/io/grpc/netty/WriteQueue.java	76.69% <0.00%> (+2.25%)	⬆️
...va/org/tikv/common/region/StoreHealthyChecker.java	79.74% <0.00%> (+7.59%)	⬆️
src/main/java/org/tikv/common/region/TiStore.java	74.35% <0.00%> (+12.82%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 3724e87...6da3e99. Read the comment docs.

[iosmanthus]

We might add some comments about this update since the scan has some implicit side effects. In addition, this issue might occur in

client-java/src/main/java/org/tikv/common/operation/iterator/RawScanIterator.java

Line 49 in 9ce52d3

TiRegion loadCurrentRegionToCache() throws GrpcException {

, the raw scan of client-java. We might add some mock tests for this interface, at least a TODO should be added.

[TrafalgarRicardoLu]

I have add comment and fix it in RawScanIterator. I think test it is not easy, so a TODO is added.

[iosmanthus]

LGTM

[ti-srebot]

@zhangyangyu, Thanks for your review. The bot only counts LGTMs from Reviewers and higher roles, but you're still welcome to leave your comments. You are not a reviewer or committer or co-leader or leader.

[sunxiaoguang]

Thanks for your contribution!

Could you please add test code to pass the code coverage check, thanks.

[iosmanthus]

LGTM

[iosmanthus]

Thanks for your contribution!

Could you please add test code to pass the code coverage check, thanks.

We may add some mock tests in the next pull request.

[ti-srebot]

cherry pick to release-3.3 in PR #625