Helm install dgraph lambda failed to pull image "dgraph/dgraph-lambda:v1.2.4"

Helm install failed to pull image “dgraph/dgraph-lambda:v1.2.4”.
I’m following the steps of https://github.com/dgraph-io/charts/tree/master/charts/dgraph-lambda to deploy Dgraph cluster and Lambda resolver. Though dgraph cluster runs successfully the lambda resolve failed.

NAME READY STATUS RESTARTS AGE
my-lambda-dgraph-lambda-f5798bd86-lpf7n 0/1 ImagePullBackOff 0 10m
my-release-dgraph-alpha-0 1/1 Running 0 99m
my-release-dgraph-alpha-1 1/1 Running 0 98m
my-release-dgraph-alpha-2 1/1 Running 0 98m
my-release-dgraph-zero-0 1/1 Running 0 99m
my-release-dgraph-zero-1 1/1 Running 0 98m
my-release-dgraph-zero-2 1/1 Running 0 98m

Here is the events in the log of pod my-lambda-dgraph-lambda-f5798bd86-lpf7n:

Events:
Type Reason Age From Message


Normal Scheduled 11m default-scheduler Successfully assigned default/my-lambda-dgraph-lambda-f5798bd86-lpf7n to ip-192-168-20-162.us-west-1.compute.internal
Normal Pulling 9m33s (x4 over 11m) kubelet Pulling image “dgraph/dgraph-lambda:v1.2.4”
Warning Failed 9m32s (x4 over 11m) kubelet Failed to pull image “dgraph/dgraph-lambda:v1.2.4”: rpc error: code = Unknown desc = Error response from daemon: manifest for dgraph/dgraph-lambda:v1.2.4 not found: manifest unknown: manifest unknown
Warning Failed 9m32s (x4 over 11m) kubelet Error: ErrImagePull
Warning Failed 9m18s (x6 over 11m) kubelet Error: ImagePullBackOff
Normal BackOff 64s (x40 over 11m) kubelet Back-off pulling image “dgraph/dgraph-lambda:v1.2.4”

1 Like

Confirmed. This is an issue, solution below.

Root Cause, the images were versioned with vX.Y.Z, but then starting with 1.2.4, they are not versioned as X.Y.Z.

Unfortunately, this is not consistent, and the chart, when the chart was updated, I should have tested this fully. I didn’t really the version scheme was changed. Grr.

Workaround:

helm install \
  my-dgraph-lambda \
  dgraph/dgraph-lambda \
  --set image.tag=1.2.4
2 Likes

Thanks for quick reply Joaquin. I just found another issue, I’m not sure if this is the proper place to describe it or not, for now I just post it here.
I cleared up everything and redid the steps following the guidance and the above solution, and then found a new bug, please check http://discuss.dgraph.io/t/dgraph-lambda-failed-to-mutate-query-data/17765

@Ann_Zhang,

The 400 error from a lambda is usually an indication that there’s an error in the JS. If you have access to the lambda container logs, can you share any errors that you find in there?

Sorry, I moved the issue to another topic http://discuss.dgraph.io/t/dgraph-lambda-failed-to-mutate-query-data/17765

Can we discuss over there? I’ll paste the logs of lambda pod over there.