• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Component/s: jclouds-plugin
• Labels:

  • openstack
  • plugin
  • slave
  • windows
• Environment:

  Open Stack
  Jenkins 1613

1. 

   10.png

   11 kB

   2015-05-15 18:44

2. 

   9.png

   21 kB

   2015-05-15 17:30

3. 

   5.png

   11 kB

   2015-05-15 15:11

4. 

   4.png

   17 kB

   2015-05-15 15:11

5. 

   3.png

   3 kB

   2015-05-15 15:00

6. 

   2.png

   11 kB

   2015-05-15 13:59

7. 

   3.png

   20 kB

   2015-05-15 13:59

[JENKINS-28403] OpenStack-Clouds-Plugin (single-use slave has already been used and is pending removal for 10 Mins.)

Jenkins Guru created issue - 2015-05-14 14:12

Jenkins Guru made changes - 2015-05-15 13:59

Attachment		New: 2.png [ 29792 ]
Attachment		New: 3.png [ 29793 ]
Description	Original: When Open-stack cloud plugin dynamically provisions a VM, it uses SSH-slaves plugins to install the slave agent. Similar to this case https://issues.jenkins-ci.org/browse/JENKINS-20921  but i used Cygwin as suggested by arazauci. The problem is that it takes a windows VM a little bit more time to start the SSH daemon, so the ssh-slaves plugin fails to start the SSH connection since windows did not start just yet. 5 seconds later when the VM launches, i can connect the slave node. However, i would have to do this manually to trigger a job every time. Can we add an a time out for the ssh slaves configuration in openstack provisioning ?	New: When Open-stack cloud plugin dynamically provisions a VM, it uses SSH-slaves plugins to install the slave agent. Similar to this case https://issues.jenkins-ci.org/browse/JENKINS-20921  but i used Cygwin as suggested by arazauci. Jenkins eventually manages to ssh to the Windows machine and install the agent. The weird behavior occurs the second after it Jenkins node shows "Slave successfully connected and online", another instances gets created and deleted instantly, while the one that runs the job stays running stale. OpenStack have already deleted it but the machine is actually still running under Xen Hypervisor. Jenkins still sees the VM and eventually it gets deleted. i tried checking\Unchecking single use slave and i had the same behavior. Why are 2 instances created instead of 1 ? why does it keep the VM online for so long
Priority	Original: Major [ 3 ]	New: Blocker [ 1 ]
Summary	Original: SSH-slaves fails because Windows VM did not start yet on OpenStack	New: weird behaviour openstack 2 instances are created instead of one and only one gets deleted instantly, the other takes 30 minutes

Jenkins Guru made changes - 2015-05-15 13:59

Summary

Original: weird behaviour openstack 2 instances are created instead of one and only one gets deleted instantly, the other takes 30 minutes

New: Weird behavior openstack-plugin

Jenkins Guru made changes - 2015-05-15 14:07

Description

Original: When Open-stack cloud plugin dynamically provisions a VM, it uses SSH-slaves plugins to install the slave agent. Similar to this case https://issues.jenkins-ci.org/browse/JENKINS-20921  but i used Cygwin as suggested by arazauci. Jenkins eventually manages to ssh to the Windows machine and install the agent. The weird behavior occurs the second after it Jenkins node shows "Slave successfully connected and online", another instances gets created and deleted instantly, while the one that runs the job stays running stale. OpenStack have already deleted it but the machine is actually still running under Xen Hypervisor. Jenkins still sees the VM and eventually it gets deleted. i tried checking\Unchecking single use slave and i had the same behavior. Why are 2 instances created instead of 1 ? why does it keep the VM online for so long

New: When Open-stack cloud plugin dynamically provisions a VM, it uses SSH-slaves plugins to install the slave agent. Similar to this case https://issues.jenkins-ci.org/browse/JENKINS-20921  but i used Cygwin as suggested by arazauci. Jenkins eventually manages to ssh to the Windows machine and install the agent. The weird behavior occurs the second after it Jenkins node shows "Slave successfully connected and online", another instances gets created and deleted instantly, while the one that runs the job stays running stale. Jenkins still sees the VM, I have to (delete slave) manually. i tried checking\Unchecking single use slave and i had the same behavior. Why are 2 instances created instead of 1 ? why does it keep the second one online?

Collapse comment: Jenkins Guru added a comment - 2015-05-15 14:09

Jenkins Guru added a comment - 2015-05-15 14:09

openstack-cloud-plugin

Expand comment: Jenkins Guru added a comment - 2015-05-15 14:09

Jenkins Guru added a comment - 2015-05-15 14:09 openstack-cloud-plugin

Jenkins Guru made changes - 2015-05-15 14:34

Component/s		New: jclouds-plugin [ 16529 ]
Component/s	Original: ssh-slaves-plugin [ 15578 ]

Jenkins Guru made changes - 2015-05-15 15:00

Attachment		New: 3.png [ 29795 ]
Description	Original: When Open-stack cloud plugin dynamically provisions a VM, it uses SSH-slaves plugins to install the slave agent. Similar to this case https://issues.jenkins-ci.org/browse/JENKINS-20921  but i used Cygwin as suggested by arazauci. Jenkins eventually manages to ssh to the Windows machine and install the agent. The weird behavior occurs the second after it Jenkins node shows "Slave successfully connected and online", another instances gets created and deleted instantly, while the one that runs the job stays running stale. Jenkins still sees the VM, I have to (delete slave) manually. i tried checking\Unchecking single use slave and i had the same behavior. Why are 2 instances created instead of 1 ? why does it keep the second one online?	New: When the Job finishes building. The slave status becomes "Single-use slave has already been used and is pending removal." but the VM is still stale. Force deleting it switches to "Slave is pending removal.", until it eventually gets deleted. The VM should get deleted instantly after the job finishes.
Summary	Original: Weird behavior openstack-plugin	New: OpenStack-Clouds-Plugin takes forever to delete single use slaves

Jenkins Guru made changes - 2015-05-15 15:05

Summary

Original: OpenStack-Clouds-Plugin takes forever to delete single use slaves

New: OpenStack-Clouds-Plugin (single-use slave has already been used and is pending removal)

Jenkins Guru made changes - 2015-05-15 15:11

Attachment		New: 5.png [ 29796 ]
Attachment		New: 4.png [ 29797 ]

Jenkins Guru made changes - 2015-05-15 15:58

Description

Original: When the Job finishes building. The slave status becomes "Single-use slave has already been used and is pending removal." but the VM is still stale. Force deleting it switches to "Slave is pending removal.", until it eventually gets deleted. The VM should get deleted instantly after the job finishes.

New: When the Job finishes building. The slave status becomes "Single-use slave has already been used and is pending removal." but the VM is still stale. Force deleting it switches to "Slave is pending removal.", until it eventually gets deleted. if i do not interfere by force deleting the VM, it gets deleted after 10 Mins. The VM should get deleted instantly after the job finishes.

Load more newer events