Skip to content
This repository has been archived by the owner on Sep 29, 2021. It is now read-only.

Docs improvement for rolloutOptions #1307

Open
walles opened this issue Jan 18, 2021 · 1 comment
Open

Docs improvement for rolloutOptions #1307

walles opened this issue Jan 18, 2021 · 1 comment

Comments

@walles
Copy link

walles commented Jan 18, 2021

Hi!

Have a look here:
https://github.com/spotify/helios/blob/master/docs/user_manual.md#rolloutoptions

I would like to know details about these rolloutOptions fields:

  • migrate: what does this do?
  • parallelism: number of nodes that will be updated in parallel I assume, pretty certain about this one
  • timeout: is this in seconds? Is it per node or for the whole rollout job?
  • overlap: what does this do?
  • token: where is this token used? why is it "insecure"?
  • ignoreFailures: what happens on failures with this on? Is the rollout marked as success in spite of the failures?
@walles
Copy link
Author

walles commented Jan 18, 2021

Found this, should probably be part of the user manual document:

walles@gew1-preprodbuildfarmworker-k-rzvm:~$ helios rolling-update -h
usage: helios rolling-update [-h] [-z MASTER] [-d DOMAINS] [--srv-name SRV_NAME] [-u USERNAME] [--google-credentials GOOGLE_CREDENTIALS] [-v] [--version] [--json] [-k] [--http-timeout HTTP_TIMEOUT]
              [--retry-timeout RETRY_TIMEOUT] [-t TIMEOUT] [-p PARALLELISM] [--async] [-T ROLLOUT_TIMEOUT] [--migrate] [--overlap] [--token [TOKEN]] [--ignore-failures] job deployment-group-name

positional arguments:
  job                    Job id.
  deployment-group-name  Deployment group name

optional arguments:
  -h, --help             show this help message and exit
  -t TIMEOUT, --timeout TIMEOUT
                         Fail rollout if a job takes longer than this to reach RUNNING (seconds)
  -p PARALLELISM, --par PARALLELISM
                         Number of hosts to deploy to concurrently
  --async                Don't block until rolling-update is complete (default: false)
  -T ROLLOUT_TIMEOUT, --rollout-timeout ROLLOUT_TIMEOUT
                         Exit if rolling-update takes longer than the given value (minutes). Note that this will NOT abort the rolling update, it will just cause this command to exit. (default: 60)
  --migrate              When specified a rolling-update will undeploy not only jobs previously deployed by the deployment-group but  also  jobs  with the same job id. Use it ONCE when migrating a service to using
                         deployment-groups
  --overlap              When specified a rolling-update will, for every host, first deploy the new version of a job before  undeploying the old one. Note that the command will fail if the job contains static port
                         assignments.
  --token [TOKEN]        Insecure access token meant to prevent accidental changes to your job (e.g. undeploys).
  --ignore-failures      When specified, the rolling-update will ignore *all* failures and will proceed to deploying  the  job  to  all  hosts in the deployment group. The rolling-update will go through the normal
                         rollout plan (respecting the --par and --overlap settings), and will wait for the job to reach  RUNNING on each host as normal; however, any failure that would otherwise cause the rolling-
                         update to abort and set the deployment group's status to FAILED is *ignored*. Be *VERY*  careful  about  using  this option, as it has the potential to completely take down your service by
                         rolling out a broken job to all of the hosts in your group.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant