👋 Hello!
I'm making use of the _EXPERIMENTAL_DAGGER_RUNNER_HOST environment variable to connect to a remote engine over TCP.
However, I sometimes get a weird connection reset by peer during the module load.
The dagger engine is located in a kubernetes cluster with a LoadBalancer in front of it to expose it.
I don't think I can change any setting in GKE to control the LoadBalancer. Is there anything I can configure on the client side to not get that error?
Here is the error message:
dagger call -m ssh://github.com/vmaffet/daggerverse@modules-v8 test
: connect
: [0.0s] | cloud url=https://dagger.cloud/traces/setup
: ┆ starting engine
: ┆ starting engine DONE [0.0s]
: ┆ connecting to engine
: ┆ [0.2s] | 21:18:03 INF connected name=dagger-engine-arm64-dagger-helm-engine-6bhnm client-version=v0.19.0 server-version=v0.19.2
: ┆ connecting to engine DONE [0.2s]
: ┆ starting session
: ┆ starting session DONE [0.1s]
: connect DONE [0.4s]
: load module: ssh://github.com/vmaffet/daggerverse@modules-v8
: | finding module configuration
: | finding module configuration DONE [3.0s]
: | initializing module inspecting module metadata
: | initializing module DONE [7.7s]
: | inspecting module metadata ERROR [0.0s]
: | ! Post "http://dagger/query": read tcp 172.16.39.34:45068->34.11.128.101:1234: read: connection reset by peer
: load module: ssh://github.com/vmaffet/daggerverse@modules-v8 ERROR [10.7s]
: | ! Post "http://dagger/query": read tcp 172.16.39.34:45068->34.11.128.101:1234: read: connection reset by peer