3

I am running dockerized hadoop cdh 4.6.0 with supervisor . I have opened the docker container and tried the following two commands. Got confused a lot.

 jps

 103 -- process information unavailable
 291 -- process information unavailable
 107 -- process information unavailable
 449 Jps

I read process info unavail means process is zombie. But i checked the status of these processes using in the same container

root@1389ca416997:/# /etc/init.d/hadoop-hdfs-namenode status
   * Hadoop namenode is running
root@1389ca416997:/# /etc/init.d/hadoop-hdfs-secondarynamenode status
   * Hadoop secondarynamenode is running
root@1389ca416997:/# /etc/init.d/hadoop-hdfs-datanode status
   * Hadoop datanode is running

What causes this ambiguity? or Am i wrong?

2 Answers 2

1

jps uses files in /tmp/hsperfdata_<username> to determine vmids and info. But sometimes the files are out of date (for whatever reason). Then, jps may start to report on stale vmids.

2
  • I don't know this fact. I am seeing this info unavail message after immediately starting the service
    – Gibbs
    Jun 1, 2015 at 9:07
  • Check the user permissions on the directory, for one thing. Make sure the user the target JVM is running as can write to that directory. Jun 1, 2015 at 9:35
1

I guess you have resolved this already but I write here for other people that might be still searching for an answer. The problem is probably related to this java bug here

A workaround is creating a soft link for the particular process file to a file in the format that jps expects to find. For example: ln -s /tmp/hsperfdata_hdfs/553 /tmp/hsperfdata_553

Alternatively, you could try downgrade your java version. For me this was working up to 1.7.0_65 - Maybe it works with later versions as well, but I start encountering the problem with 1.7.0_79

Also, if you don't mind checking only with the root user you could just run jps with he user that is the owner of the /tmp/hsperfdata_user file. For example: sudo -u hdfs jps and sudo -u mapred jps will display the processes running for /tmp/hsperfdata_hdfs/ and /tmp/hsperfdata_mapred/ respectively.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.