[HADOOP-11252] RPC client does not time out by default

• Type: Bug
• Status: Closed
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 2.5.0
• Fix Version/s: 2.8.0, 2.7.3, 2.6.4, 3.0.0-alpha1
• Component/s: ipc
• Target Version/s: 2.8.0, 2.6.4
• Hadoop Flags: Incompatible change, Reviewed

Release Note

This fix includes public method interface change.
A follow-up JIRA issue for this incompatibility for branch-2.7 is HADOOP-13579.

• Assignee: Masatake Iwasaki
• Reporter: Wilfred Spiegelenburg
• Created: 31/Oct/14 01:57
• Updated: 26/Feb/20 05:29
• Resolved: 04/Jan/16 05:41

Description

The RPC client has a default timeout set to 0 when no timeout is passed in. This means that the network connection created will not timeout when used to write data. The issue has shown in YARN-2578 and HDFS-4858. Timeouts for writes then fall back to the tcp level retry (configured via tcp_retries2) and timeouts between the 15-30 minutes. Which is too long for a default behaviour.

Using 0 as the default value for timeout is incorrect. We should use a sane value for the timeout and the "ipc.ping.interval" configuration value is a logical choice for it. The default behaviour should be changed from 0 to the value read for the ping interval from the Configuration.

Fixing it in common makes more sense than finding and changing all other points in the code that do not pass in a timeout.

Offending code lines:
https://github.com/apache/hadoop/blob/trunk/hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/ipc/RPC.java#L488
and
https://github.com/apache/hadoop/blob/trunk/hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/ipc/RPC.java#L350

Issue Links

• breaks: HADOOP-14958 CLONE - Fix source-level compatibility after HADOOP-11252
• breaks: HADOOP-13579 Fix source-level compatibility after HADOOP-11252
• duplicates: HADOOP-9654 IPC timeout doesn't seem to be kicking in
• is depended upon by: HADOOP-11574 Uber-JIRA: improve Hadoop network resilience & diagnostics
• is duplicated by: YARN-2578 NM does not failover timely if RM node network connection fails
• is duplicated by: YARN-5119 Timeout HA issue: Enable IPC ping for all calls by default
• is related to: YARN-2714 Localizer thread might stuck if NM is OOM
• relates to: HADOOP-12672 RPC timeout should not override IPC ping interval