Getting error on deleting workflows while using astraDB serverless with temporal

Community Support
1

Hi im noticing that completed workflows are not getting deleted even after retention period is over
with the following error
“Fail to process task”,“shard-id”:178,“address”:“10.71.19.10:7234”,“component”:“visibility-queue-processor”,“shard-id”:178,“queue-task-id”:25165913,“queue-task-visibility-timestamp”

operation DeleteWorkflowExecution encountered Provided value ONE is not allowed for Write Consistency Level (disallowed values are: [ANY, ONE, LOCAL_ONE])"

As i understand temporal uses the value provided in the config file
visibility:
driver: “cassandra”
cassandra:
disableInitialHostLookup: true
hosts: “host”
port: 29042
keyspace: temporal_visibility
user: “user”
existingSecret: “password”
replicationFactor: 3
consistency:
default:
consistency: “local_quorum”
serialConsistency: “local_serial”

So why is the delete operation ignoring this consistency level and using ONE as the value?
Note: i am using astraDB as my database provider, i understand that they do not support consistency ONE on writes but is this query expected?

I can see as per this code

	cfg config.Cassandra,
	r resolver.ServiceResolver,
	logger log.Logger,
) (*visibilityStore, error) {
	session, err := gocql.NewSession(cfg, r, logger)
	if err != nil {
		logger.Fatal("unable to initialize cassandra session", tag.Error(err))
	}

	return &visibilityStore{
		session:      session,
		lowConslevel: gocql.One,
	}, nil
}

Thanks
3

Thanks for sharing this. Asked server team to provide input and will get back to you asap.

4

Thanks for the ack! Awaiting response

5

plus one on this. We just started testing temporal on astra, and running into this issue.

6

Thanks for reporting this. It does seem like a bug that we use consistency One for delete.

7

Hi

@Yimin_Chen Need your support to close this as this feature blocks a lot of testing.

8

Fix is landed to master.

1 Like
9

Just to add for tracking - commit here.