Providing wrong config when creating a replication group via the Console leaves the replication group "stuck" and unusable

Axon Server
1

Using Axon Server 2024.0.0, if I perform the following steps:

  1. Console → Contexts
  2. Create Context
  3. Enter context name (“test”)
  4. Replication group → Create new
  5. Add replication group name (“test”)
  6. Choose nodes (I picked a single node in my case)
  7. Roles → messaging_only

Then the creation of the context will fail because no nodes in the replication group have the PRIMARY role.

If, after this fails, I go to the Replication Groups tab, I do not see the above replication group. So far so good. However, if I go to Add Replication Group and perform the following steps:

  1. Enter same name for the replication group as in step 5 above.
  2. Choose the same node as in step 6 above (I only have the one node, so I cannot say if this occurs with multiple/other nodes).
  3. Role → PRIMARY

Then I see the following error message:

{
    "message": "[AXONIQ-2107] The creation of the replication group is already in progress.",
    "error": "REPLICATION_GROUP_UPDATE_IN_PROGRESS",
    "status": "AXONIQ-2107"
}

image
image764×213 15.3 KB

It seems that this causes the replication group creation process to get “stuck”. I also don’t see a way to fix state this via the Console.

2

Hi, on Monitoring page, in the running task do you see a task for creating this rep group? If yes, try to cancel this task and create replication group again.

Thanks for reporting this bug

3

Thanks for the quick reply.

After some time, it seems that the replication group creation process suffers a timeout (it seems around 30m or so?) and I am able re-create the replication group. So the workaround here right now would be to just grab a cup of coffee it seems. :slight_smile:

Monitoring page, in the running task do you see a task for creating this rep group?

After reproducing the issue, I do not see any running tasks in the Monitoring page:

image
image1418×248 4.69 KB

Thanks for reporting this bug

No problem - thanks for an amazing product! Looking forward to all the things to come here.

4

I reproduced it as well, and another workaround is to delete affected replication group via CLI or API for example

curl -X 'DELETE' \
  'http://localhost:8024/v1/replicationgroups/bug-rep?preserveEventStore=false' \
  -H 'accept: */*'

Anyway we will address issue in next patch…