ENTRY

"Running Selenium test suite on new Hudson Slave fails with “killed”"

Date:June 30th, 2014
Tags:
Comments: 0
  • Join date: 11-30-99
  • Posts: 23
View Counter:
  • 51 views
Rate it
Ad
Poll
  • Which virtualization do you like most?

    View Results

    Loading ... Loading ...
Feeds:
  • bodytext bodytext bodytext

I spent some hours finding the error in Hudson setup, which was not there. Conclusion was, that selenium test suite required much more system resources (memory). And as it tried to get all the memory it wanted, it was killed by the kernel. Solution was to add the memory: 516 MB -> 4146 MB. That would solve the problem.

I had recently asked our operations support division to setup one more Hudson slave system.
Well, they give hardware and I make the multi-slave Hudson system with the script I wrote.

The problem was only, that any builds on new slave never finished with success, or even any kind of reason why and what fails.

From the Hudson side I see last entries in log:

[htmlpublisher] Archiving at PROJECT level /var/log/hudson/hudson-slave-1/workspace/my.base.project/my.project/target/surefire-reports/html/ to /opt/hudson/.hudson/jobs/my.base.project/htmlreports/Report_(reportng)
ERROR: Directory '/var/log/hudson/hudson-slave-1/workspace/my.base.project/my.project/target/surefire-reports/html/' exists but failed copying to '/opt/hudson/.hudson/jobs/my.base.project/htmlreports/Report_(reportng)'.
[8mha:AAAAWB+LCAAAAAAAAABb85aBtbiIQSmjNKU4P08vOT+vOD8nVc8DzHWtSE4tKMnMz/PLL0ldFVf2c+b/lb5MDAwVRQxSaBqcITRIIQMEMIIUFgAAckCEiWAAAAA=[0mFinished: FAILURE

 
That might be the place to start…

After some hours of work I find nothing special, no configuration problem with the htmlpublisher.

Next thing I tried was to run the project on the slave itself. The result was the same output, and additional line: “Killed

Hm… Who or what would kill my build?

less /var/log/syslog gave me:

May 26 18:34:12 hudson-slave-1 kernel: [1835424.684067] Free swap = 0kB
May 26 18:34:12 hudson-slave-1 kernel: [1835424.684855] Total swap = 0kB
May 26 18:34:12 hudson-slave-1 kernel: [1835424.688070] 131072 pages of RAM
May 26 18:34:12 hudson-slave-1 kernel: [1835424.688850] 0 pages of HIGHMEM
May 26 18:34:12 hudson-slave-1 kernel: [1835424.689582] 2061 reserved pages
May 26 18:34:12 hudson-slave-1 kernel: [1835424.690303] 738 pages shared
May 26 18:34:12 hudson-slave-1 kernel: [1835424.691023] 0 pages swap cached
May 26 18:34:12 hudson-slave-1 kernel: [1835424.691749] 0 pages dirty
May 26 18:34:12 hudson-slave-1 kernel: [1835424.692197] 0 pages writeback
May 26 18:34:12 hudson-slave-1 kernel: [1835424.692948] 12 pages mapped
May 26 18:34:12 hudson-slave-1 kernel: [1835424.693660] 2750 pages slab
May 26 18:34:12 hudson-slave-1 kernel: [1835424.694359] 763 pages pagetables
May 26 18:34:12 hudson-slave-1 kernel: [1835424.695106] Out of memory: kill process 19365 (java) score 266457 or a child
May 26 18:34:12 hudson-slave-1 kernel: [1835424.696652] Killed process 19366 (firefox)

 
Next, I compared the system memory on other slaves and on this one. This slave has only 516 MB, and others – 4146 MB (with the help of hardinfo)

That was it!

👉 Selenium needs much more memory then that to run. But Hudson was fine and tests were also ok.

Be Sociable, Share!
 
Interesting entry?
VN:F [1.8.5_1061]
Rating: 0 (from 0 votes)
0 votes 'YES'  0 votes 'NO'


TAGS
No tags for this post.

RELATED
Pages
Posts

SOCIAL
Be Sociable, Share!

INCOMING SEARCH TERMS



Responces

Leave a Reply