Closed ddgenome closed 5 years ago
@cdupuis I see this in the logs:
2019-08-13T11:22:37.996Z [m:10] [debug] Mutating 'https://automation.atomist.com/graphql/team/T095SFFBK' with variables '{"name":"SyncRepoCommand","description":"Executing `SyncRepoCommand`","owner":"@atomist/k8s-sdm_k8s-internal-staging","data":"{\"secrets\":[]}","tasks":[]}' and mutation: mutation CreateJob( $data:
String! $name: String! $description: String! $owner: String! $tasks: [AtmJobTaskInput!]!) { createAtmJob( jobInput: { data: $data name: $name description: $description owner: $owner jobTasks: $tasks } ) { id }}
…
2019-08-13T11:22:38.312Z [m:10] [debug] Mutation returned data: {"createAtmJob":{"id":"8ddb7dbc-e1ac-4408-9088-235b142a7562"}}
But that id does not appear anywhere else in the logs and I see no logs of the sync ever occurring.
For Jobs to work, you need the job support extension pack in your SDM. That gets installed internally but maybe not in your k8s-sdm. Do you see it listed in the start up message?
I’m talking about https://github.com/atomist/sdm-core/blob/master/lib/pack/job/job.ts#L29
That's probably it, I'll add that.
Is it acceptable to call jobSupport()
from a StartupListener, or is that something that has to be handled in configuration post-processors? And is it bad if jobSupport()
is called more than once?
Calling jobSupport()
is not necessary, that is added by default to all SDMs.
I did find what appears to be a bug around startupListeners and triggeredListeners, atomist/sdm#762 .
Wrap sync repo functionality in a command. Use command to run sync of entire repo as a job so it gets scheduled in a worker.
Fix naming of kube undeploy.
@cdupuis do you think this would work?