[HBASE-10813] Possible over-catch of exceptions

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.96.1
• Fix Version/s: 0.99.0
• Component/s: regionserver, util
• Hadoop Flags: Reviewed
• Assignee: Ding Yuan
• Reporter: Ding Yuan
• Created: 24/Mar/14 02:03
• Updated: 21/Feb/15 23:28
• Resolved: 09/Apr/14 16:13

Description

There are a few cases found by a tool that are possibly over-catch of exceptions, especially those that will abort the server. Over-catching these exceptions may unexpectedly abort the server, and may cause problems in the future when code in the try-block evolves. I am attaching a patch against trunk that constrains the catch blocks to the exact exceptions that were thrown.

My tool actually found one more case in 0.96.1, but I found it has already been fixed in trunk:

Line: 1175, File: "org/apache/hadoop/hbase/master/SplitLogManager.java"
1173:             try {
1174:               Thread.sleep(20);
1175:-            } catch (Exception ignoreE) {
1175:+           } catch (InterruptedException e) {
1176:               // ignore
1177:             }

Any feedbacks are much appreciated!

Issue Links

• relates to: HBASE-10954 Fix TestCloseRegionHandler.testFailedFlushAborts