[IOTDB-5620] Fix flush stuck when there is a lot of time partitions in each DataRegion

[HTHou]

Description

When set Max_heap_size=2G and insert data of 7 years. The flush thread will be stuck.

There is a simple way to reproduce.

Set max_heap_size of data node to 200M, set write_memory_proportion to 10000000:1.

Then start standalone iotdb and execute the following sqls:

insert into root.sg.d1(time,s1) values(1,2);
flush;
insert into root.sg.d1(time,s1) values(2,2);
insert into root.sg.d1(time,s1) values(604800001,2);
flush;

Finally, you will see the last flush cannot execute successfully.

Reason

The evictOldPartition method in TimePartitionManager always peek the first TimePartitionInfo in the TreeSet in the while loop. However, the memSize of the TimePartitionInfo may be 0. In this case, the loop won't be ended.

Solution

Change the treeSet.first() to treeSet.pollFirst().

[sonarqubecloud]

Kudos, SonarCloud Quality Gate passed!   

0 Bugs
0 Vulnerabilities
0 Security Hotspots
1 Code Smell

0.0% Coverage
0.0% Duplication