Make Lando not die on startup failures, loop and keep trying
Categories
(Conduit :: Lando, enhancement, P1)
Tracking
(Not tracked)
People
(Reporter: smacleod, Assigned: smacleod)
Details
(Keywords: conduit-story, conduit-triaged)
Attachments
(1 file)
This is important for running in GCP.
|
Assignee | |
Comment 1•6 years ago
|
||
This change consolidates loading configuration from the environment and
then initializing and readying various subsystems. These new ready
checks will be re-run as they fail making problems easier to diagnose in
kubernetes and service startup more resiliant. A couple of ready checks
will block startup due to misconfiguration of important environment
variables.
Some subsystems that should have a ready check, such as the database,
don't. This was chosen to simplify the change and maintain parity with
the previous version. Subsystems without a ready check should have an
appropriate health check and will be included in a /heartbeat
response.
As an added bonus the celery worker process will also go through the
same ready checks and has proper logging in MozLog format. This should
make diagnosing startup issues with the worker a little easier. As part
of testing a celery sub-command was added to lando-cli which should
allow running any of the celery commands with proper lando
configuration.
|
|
Assignee | |
Updated•6 years ago
|
Description
•