[JENKINS-734] clock NPE for jnlp slaves

[jenkins-infra-bot]

I believe this happens if the 'launch command' is left blank. That's the case
when configuring a slave for jnlp access only.

---

Originally reported by arnaud_lacour, imported from: clock NPE for jnlp slaves

• assignee: kohsuke
• status: Closed
• priority: Major
• component(s): remoting
• resolution: Fixed
• resolved: 2008-01-20T13:32:59+00:00
• votes: 0
• watchers: 0
• imported: 2025-11-25

Raw content of original issue

I believe this happens if the 'launch command' is left blank. That's the case
when configuring a slave for jnlp access only.

• environment: Platform: All, OS: All

1 attachment

• hudson-clock-NPE.png

  

[jenkins-infra-bot]

arnaud_lacour:

• Original comment link
• Raw content of original comment:

  Created an attachment (id=80)
  clock NPE for jnlp slave
  

Created an attachment (id=80)
clock NPE for jnlp slave

[jenkins-infra-bot]

super_kohsuke:

• Original comment link
• Raw content of original comment:

  The line numbers don't match up. Which version of Hudson is this?
  

The line numbers don't match up. Which version of Hudson is this?

[jenkins-infra-bot]

arnaud_lacour:

• Original comment link
• Raw content of original comment:

  1.121 I believe.
  I upgraded to 1.129 this morning and it still occurs.
  I figured it's not because of an empty field but when the slave isn't connected.
  When the slave is connected it  does not happen even with empty config values.
  

1.121 I believe.
I upgraded to 1.129 this morning and it still occurs.
I figured it's not because of an empty field but when the slave isn't connected.
When the slave is connected it does not happen even with empty config values.

[jenkins-infra-bot]

kohsuke:

• Original comment link
• Raw content of original comment:

  Any chance you can get me a new stack trace?
  

Any chance you can get me a new stack trace?

[jenkins-infra-bot]

kohsuke:

• Original comment link
• Raw content of original comment:

  Closing this as I can't reproduce the problem any more as of 1.134. This must
  have been fixed since then.
  

Closing this as I can't reproduce the problem any more as of 1.134. This must
have been fixed since then.

[jenkins-infra-bot]

tblack:

• Original comment link
• Raw content of original comment:

  reopen, as the issue is reproducible on v1.145
  

reopen, as the issue is reproducible on v1.145

[jenkins-infra-bot]

tblack:

• Original comment link
• Raw content of original comment:

  I traced the problem with remote debugging as detailed on the wiki,
  http://hudson.gotdns.com/wiki/display/JENKINS/Building+Hudson. maven + jetty rocks!
  
  The problem occurs when you leave the "# of executors" field blank.  This causes
  the slave to never be added to the computer map (Hudson.java, line 451).
  Additionally, the slave will not show up on the Build Executor Pane.  Upon
  reloading the System Configuration page, the slave config will be displayed,
  immediately followed by:
  
  Status Code: 500
  Exception:
  Stacktrace:
  
  java.lang.NullPointerException
  	at hudson.model.Slave.getClockDifference(Slave.java:205)
  ...
  
  So, I understand the reason behind setting the executors to zero (according to
  the help text for that field), but I think for the initial new slave config,
  that field should be mandatory > 0.
  
  If this is not reasonable, then maybe handle the NPE with an appropriate message
  along the lines of "This slave is temporarily disabled due to the Number of
  Executors being zero"
  

I traced the problem with remote debugging as detailed on the wiki,
http://hudson.gotdns.com/wiki/display/JENKINS/Building+Hudson. maven + jetty rocks!

The problem occurs when you leave the "# of executors" field blank. This causes
the slave to never be added to the computer map (Hudson.java, line 451).
Additionally, the slave will not show up on the Build Executor Pane. Upon
reloading the System Configuration page, the slave config will be displayed,
immediately followed by:

Status Code: 500
Exception:
Stacktrace:

java.lang.NullPointerException
at hudson.model.Slave.getClockDifference(Slave.java:205)
...

So, I understand the reason behind setting the executors to zero (according to
the help text for that field), but I think for the initial new slave config,
that field should be mandatory > 0.

If this is not reasonable, then maybe handle the NPE with an appropriate message
along the lines of "This slave is temporarily disabled due to the Number of
Executors being zero"

[jenkins-infra-bot]

kohsuke:

• Original comment link
• Raw content of original comment:

  Fixed in 1.175.
  

Fixed in 1.175.

[jenkins-infra-bot]

akostadinov:

• Original comment link
• Raw content of original comment:

  Will that break a setup where hudson master has 0 executors? That's useful to
  ensure master is not used for any jobs.
  
  Thanks.
  

Will that break a setup where hudson master has 0 executors? That's useful to
ensure master is not used for any jobs.

Thanks.