You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We have some projects which have a lot of dependencies (25+ with transitive dependencies, usually one deployment per dependency), and the initial devspace deploy can take quite a bit of time (ie. a few minutes rather than less than a minute - which is what we had with our docker-compose setup). Once everything is running, devspace dev -x is working great!
Part of this increased spin-up time may be due to this: we seem to sometimes end up with the same dependencies being deployed multiple times... e.g as part of the mentioned dependency graph we have multiple services depending on the same postgres dependency (same git source/branch/vars etc.), but I still see the postgres helm chart being installed/upgraded multiple times in the logs:
I expect postgres to only be helm install-ed once, ie. the helm revision for the postgres release should be 1. (The output above is only a simple example, in our case some helm releases seem to be upgraded 10+ times).
How can we reproduce the bug? (as minimally and precisely as possible)
The addition of the POSTGRES_USER variable to the postgres dependency seems to be what causes the issue (bforchhammer/devspace-dependency-example@0fe3258); without it the postgres helm release is only installed once!
Local Environment:
DevSpace Version: 5.18.1
Operating System: mac
ARCH of the OS: AMD64 Kubernetes Cluster:
Cloud Provider: docker-desktop
Kubernetes Version: v1.22.4
Anything else we need to know?
Happy to provide more info if needed :)
/kind bug
The text was updated successfully, but these errors were encountered:
@bforchhammer thanks for creating this issue! Yes you are correct, DevSpace shouldn't deploy the postgres dependency twice. We'll fix that in the next release!
What happened?
We have some projects which have a lot of dependencies (25+ with transitive dependencies, usually one deployment per dependency), and the initial
devspace deploy
can take quite a bit of time (ie. a few minutes rather than less than a minute - which is what we had with our docker-compose setup). Once everything is running,devspace dev -x
is working great!Part of this increased spin-up time may be due to this: we seem to sometimes end up with the same dependencies being deployed multiple times... e.g as part of the mentioned dependency graph we have multiple services depending on the same
postgres
dependency (same git source/branch/vars etc.), but I still see thepostgres
helm chart being installed/upgraded multiple times in the logs:What did you expect to happen instead?
I expect
postgres
to only behelm install
-ed once, ie. the helm revision for thepostgres
release should be1
. (The output above is only a simple example, in our case some helm releases seem to be upgraded 10+ times).How can we reproduce the bug? (as minimally and precisely as possible)
I created an example to reproduce the issue here: https://github.com/bforchhammer/devspace-dependency-example
The addition of the
POSTGRES_USER
variable to thepostgres
dependency seems to be what causes the issue (bforchhammer/devspace-dependency-example@0fe3258); without it thepostgres
helm release is only installed once!Local Environment:
Kubernetes Cluster:
Anything else we need to know?
Happy to provide more info if needed :)
/kind bug
The text was updated successfully, but these errors were encountered: