[AMBARI-2088] Cluster installation times out at server side too fast

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.2.3
• Fix Version/s: 1.2.3
• Component/s: ambari-server
• Assignee: Siddharth Wagle
• Reporter: Siddharth Wagle
• Created: 07/May/13 17:56
• Updated: 08/May/13 05:46
• Resolved: 07/May/13 18:04

Description

Cluster installation fails on Datanode with task timeout aproximately in 5 minutes. It is not an UI issue, I've checked via API that server considers the request status to be TIMEDOUT, and agent continues running puppet manifests (each resulting with 0 return code). Agent logs does not contain watchdog messages. Our local internet connection is slower then Amazon internal network, and as a result installation takes more time (in my case, this issue reproduces almost every time when installing on a 2-node cluster). Clicking retry results in successful installation due to cached packages on previous attempt.

We have different timeout values at agent and server (10 minutes and 5 minutes).
/src/main/python/ambari_agent/PuppetExecutor.py:42
PUPPET_TIMEOUT_SECONDS = 600
com.google.inject.AbstractModule#bindConstant
bindConstant().annotatedWith(Names.named("actionTimeout")).to(300000L);