2

I have a docker container that I can excecute fine locally. Yet when run on a mesos cluster, I get SIGTERMS

/usr/my_script.sh: line 57:   310 Killed xsltproc sort.xsl ${2} > ${2}_bat
W0703 09:09:54.465442  5074 logging.cpp:91] RAW: Received signal SIGTERM from process 2262 of user 0; exiting

I don't understand where this problem is comming from and how to best debug it. How can I find out what's killing my container?

I tried increasing the RAM of the container available to over 4GB, yet to no avail. Furthermore, according to /usr/bin/time -v xsltproc sort.xsl offending_file.xml > sortedFile.xml the process should only consume 1GB RAM.

I also tried googling for the error output of W0703 and 5074 logging.cpp:91, yet to no avail. It also begs the question why the container has no problem executing the command when run locally.

k0pernikus
  • 60,309
  • 67
  • 216
  • 347
  • 1
    How much memory is available on the host? Can you check the host `dmesg` output to see if the OOM killer might have killed it? The `Killed` in the log output is suspect of the OOM killer. – Andy Shinn Jul 03 '17 at 18:00
  • Similar to [this question](https://serverfault.com/questions/858982/who-is-killing-my-docker-container). Can you post more details? Logs from mesos-agent and executor. Are you usinge Mesos Healthchecks? – janisz Jul 04 '17 at 06:45

1 Answers1

2

I had this same issue. I was running a docker container on Chronos and left the "command" field blank, assuming it would execute CMD in the Dockerfile when not overridden. Explicitly copying the command into the Mesos configuration fixed the issue for me.

Grant Palmer
  • 208
  • 2
  • 7