#Multiple clients hitting the same engine resulting in what looks like incorrect cache hits

1 messages · Page 1 of 1 (latest)

weak kraken
#

Gitlab matrix jobs -> multiple jobs hitting the client at the same time. Some jobs succeed, others fail with a withExec error that clearly contains some other job's information. This looks like cache hits that shouldn't be cache hits.

Think a short 5m debugging call will be easiest to show this. I'm hoping the problem is in my function caching and/or how I've built the module rather than a Dagger engine problem!

fiery patrol
weak kraken
#

I'll add the token tomorrow morning and share some traces.

fiery patrol
#

👂

weak kraken
#

@fiery patrol

Had a workshop this morning, got round to this now. Some examples coming up. In each case, the WithConfig method will have a TF_VAR_account_id and TF_VAR_environment, the values of which will be in the terraform commands. In each case, the terraform init works, the terraform plan tries to reach some other account in the matrix job.

  1. Failed
    https://dagger.cloud/mjb/traces/e0d8703d9e2af096f379299851e2fa64 Account ends 199 environment dv, error trying to reach account ending 892 environment at)

  2. Success
    https://dagger.cloud/mjb/traces/663ccb4d5b915cb9e132dbf31a0250dc This is the account ending 892 environment at from the previous error. This job was in the same gitlab pipeline, ran at the same time, and succeeded.