[HBASE-17131] Avoid livelock caused by HRegion#processRowsWithLocks

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.4.0, 1.3.1, 1.2.6, 2.0.0
• Fix Version/s: 1.4.0, 1.3.2, 2.0.0, 1.2.7
• Component/s: regionserver
• Hadoop Flags: Reviewed
• Assignee: Chia-Ping Tsai
• Reporter: Chia-Ping Tsai
• Created: 20/Nov/16 09:58
• Updated: 28/Jul/17 09:41
• Resolved: 28/Jul/17 06:09

Description

HRegion.java

try {
      // STEP 2. Acquire the row lock(s)
      acquiredRowLocks = new ArrayList<RowLock>(rowsToLock.size());
      for (byte[] row : rowsToLock) {
        // Attempt to lock all involved rows, throw if any lock times out
        // use a writer lock for mixed reads and writes
        acquiredRowLocks.add(getRowLockInternal(row, false));
      }
      // STEP 3. Region lock
      lock(this.updatesLock.readLock(), acquiredRowLocks.size() == 0 ? 1 : acquiredRowLocks.size());
      locked = true;
      boolean success = false;
      long now = EnvironmentEdgeManager.currentTime();
      try {

We should lock all involved rows in the second try-finally. Otherwise, we won’t release the previous locks if any subsequent lock times out.

Issue Links

• is related to: HBASE-18464 Backport HBASE-17131 (Avoid livelock caused by HRegion#processRowsWithLocks) into branch-1.3 and branch-1.2