No Containers being spawned

[djl197]

All,

Just started to play with this.
I have a Jenkins Master with the plugin installed.
I set it up to point at another machine as a docker host.
I run Test Connection and it reports version 1.0.1.

I setup a cloud with the ID set to:
localhost:5000/jenkins_slave

On the remote machine - This can be seen (docker images)
I setup the access, home dir etc and limit it to 8 instances.

I have a single slave setup, and I was expecting that any subsequent jobs would trigger the creation of a docker container.
However this is not happening.

Under the 'System Configuration->Docker' it shows the image is available with no running containers.

I have enabled debug (System Logs)
com.nirima.jenkins.plugins.docker Set to ALL

But this just confirms connections are made:
Jul 04, 2014 11:32:45 AM INFO com.nirima.jenkins.plugins.docker.DockerCloud connect
Building connection to docker host Grass URL http://grass.DOMAIN:4243
Jul 04, 2014 11:32:46 AM INFO com.nirima.jenkins.plugins.docker.DockerCloud connect
Building connection to docker host Grass URL http://grass.DOMAIN:4243
Jul 04, 2014 11:32:46 AM INFO com.nirima.jenkins.plugins.docker.DockerCloud connect
Building connection to docker host Grass URL http://grass.DOMAIN:4243

Any ideas on why nothing much seems to be happening - and why the debug is not very verbose?

[thomassuckow]

Sounds a lot like #50

[thomassuckow]

Are you still having problems with this? I believe this is actually Jenkins "NodeProvisioner" deciding there is not enough work to be done to justify spinning up a cloud slave. Make sure the build has completely succeeded/failed once and see if the next time it will spin up Docker slaves.

[djl197]

Yes - if I specify a specific label then Jenkins spins up my container.
If I use the same labels as everything else it seems that Jenkins is very reluctant to spin up a container.
Even sometimes appearing to refuse to spin one up even when all 'physical' slaves are busy.

[thomassuckow]

I am curious what the queue length looks like in the Load Statistics screen. Based on what I have read, Jenkins should try to spin up a slave if it is greater than 1, but it is also dampened to prevent spinning up slaves for a single job if it can avoid it.

If Jenkins does want to spin one up and the docker plugin says it has a compatible container, it will be logged as "Excess workload" just before attempting to start a container (which may fail if it is at the limit).

[KostyaSha]

Is it still actual?

[KostyaSha]

Closed by mistake 😃 but feel free to reopen if issue is still valid