Ambari Agent Start Failed

[#1]

Environmennt: CentOS7

I manually installed the ambari-agent older version (2.2.2.0) and found out it's the wrong version I needed. Then I ran yum to remove it. I updated the repo so I now downloaded and installed the correct version 2.4.0.1.

However, when I ran "ambari-agent start", it showed the following error:

Verifying Python version compatibility...
Using python  /usr/bin/pythonChecking for previously running Ambari Agent...
Starting ambari-agentVerifying ambari-agent process status...
ERROR: ambari-agent start failed. For more details, see /var/log/ambari-agent/ambari-agent.out:
====================
Traceback (most recent call last):  
   File "/usr/lib/python2.6/site-packages/ambari_agent/AmbariAgent.py", line 24, in <module>  from Controller import AGENT_AUTO_RESTART_EXIT_CODE  
   File "/usr/lib/python2.6/site-packages/ambari_agent/Controller.py", line 41, in <module>  from ambari_agent.ActionQueue import ActionQueue  
   File "/usr/lib/python2.6/site-packages/ambari_agent/ActionQueue.py", line 37, in <module>  from ambari_commons.str_utils import split_on_chunks
ImportError: cannot import name split_on_chunks
====================

The /var/log/ambari-agent/ambari-agent.out shows the same error above.

Seems that I can't cleanly remove the dependencies that installed with the previous version. Even when I tried ambari-agent reset, the same error happened and said "ImportError: cannot import name split on chunks".

How to resolve this problem?

Many thanks!

[#2]

@William Li do the following, review the errors in /var/log/ambari-agent then if nothing is obvious go with next steps below

yum erase ambari-agent

rm -rf /var/lib/ambari-agent
rm -rf /var/log/ambari-agent
rm -rf /usr/lib/python2.6/site-packages/ambari_agent
yum install -y ambari-agent

there may be other residual directories to remove but once you clean all, reinstall and try again.

[#3]

Thanks a lot Artem for the quick response. Tried your steps and still has the exact error above. However, this time it has one extract line of error message:

  File "/usr/lib/python2.6/site-packages/ambari_agent/HeartbeatHandlers.py", line 21, in <module>
    from ambari_commons.exceptions import FatalException 

ImportError: No module named ambari_commons.exceptions

Checked the logs and there are nothing else other than this above error.

[#4]

Please see my comment below to remove the entire site-packages. After that, it worked!

[#5 · Accepted solution]

I did one more thing to

rm -rf /usr/lib/python2.6/site-packages

yum remove ambari-agent
yum install ambari-agent -y 

It worked. The ambari-agent start completed successfully.

[#6]

Great job!

[#7]

You should not do "rm -rf /usr/lib/python2.6/site-packages" as this affects to other python scripts, such as "yum"

This issue would be caused if you install ambari-server or /usr/lib/python2.6/site-packages/ambari_commons/str_utils.py is not a right one.

Please make sure "/usr/lib/python2.6/site-packages/ambari_commons" is a symlink to /usr/lib/ambari-agent/lib/ambari_commons

[#8]

I had same problem on other than master nodes, you have to update ambari-server on each node even if you are not running server on each one.