[HBASE-19709] Guard against a ThreadPool size of 0 in CleanerChore

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Fix Version/s: 2.0.0-beta-1, 2.0.0
• Hadoop Flags: Reviewed
• Assignee: Josh Elser
• Reporter: Siddharth Wagle
• Created: 04/Jan/18 22:39
• Updated: 01/Aug/18 06:22
• Resolved: 05/Jan/18 23:52

Description

Post HBASE-18309, we choose the number of threads by the following logic:

+  /**
+   * If it is an integer and >= 1, it would be the size;
+   * if 0.0 < size <= 1.0, size would be available processors * size.
+   * Pay attention that 1.0 is different from 1, former indicates it will use 100% of cores,
+   * while latter will use only 1 thread for chore to scan dir.
+   */

swagle has found on his VM that despite having two virtual processors, Runtime.getRuntime().availableProcessors() returns 0, which results in 0 threads for the pool which throws an exception.

java.lang.IllegalArgumentException
        at java.util.concurrent.ForkJoinPool.checkParallelism(ForkJoinPool.java:2546)
        at java.util.concurrent.ForkJoinPool.<init>(ForkJoinPool.java:2536)
        at java.util.concurrent.ForkJoinPool.<init>(ForkJoinPool.java:2505)
        at org.apache.hadoop.hbase.master.cleaner.CleanerChore.<init>(CleanerChore.java:112)
        at org.apache.hadoop.hbase.master.cleaner.CleanerChore.<init>(CleanerChore.java:83)
        at org.apache.hadoop.hbase.master.cleaner.LogCleaner.<init>(LogCleaner.java:65)
        at org.apache.hadoop.hbase.master.HMaster.startServiceThreads(HMaster.java:1130)
        at org.apache.hadoop.hbase.master.HMaster.finishActiveMasterInitialization(HMaster.java:813)
        at org.apache.hadoop.hbase.master.HMaster.access$500(HMaster.java:223)
        at org.apache.hadoop.hbase.master.HMaster$4.run(HMaster.java:2016)
        at java.lang.Thread.run(Thread.java:745)

We should make sure that we take the max of 1 and the computed number of threads.

Issue Links

• is related to: HBASE-18309 Support multi threads in CleanerChore