You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@artemis.apache.org by He...@systema.com on 2026/06/10 13:01:21 UTC
Is there a way, to make the clients rebalance on Artemis Cluster?
Hello team,
we are running three artemis brokers A,B,C in an "all active"
configuration.
The clients use failover transport (A,B,C)?failoverAttempts=-1 and when
the clients startup they distribute on the brokers just fine.
When broker A is restarted, clients fall to the other brokers. Still looks
good so far.
If broker B is restarted, the clients do not go back to A they all end up
on C.
Having all clients on one broker is not loadbalanced and restarting the
clents is not an option.
Did we configure something wrong? would it help to start with different
sequences in the url (A,B,C), (B,A,C) etc. ?
Regards
Herbert
Re: Is there a way, to make the clients rebalance on Artemis Cluster?
Posted by Justin Bertram <jb...@apache.org>.
It doesn't sound like you're doing anything wrong. However, it does sound
like you might want a connection router [1] configured on each broker in
the cluster to help keep things balanced.
Other things to keep in mind:
- The management API provides the ability to close client connections,
forcing them to reconnect.
- Your cluster might be too large [2].
Justin
[1]
https://artemis.apache.org/components/artemis/documentation/latest/connection-routers.html
[2]
https://artemis.apache.org/components/artemis/documentation/latest/clusters.html#performance-considerations
On Wed, Jun 10, 2026 at 8:01 AM <He...@systema.com> wrote:
> Hello team,
>
> we are running three artemis brokers A,B,C in an "all active"
> configuration.
> The clients use failover transport (A,B,C)?failoverAttempts=-1 and when
> the clients startup they distribute on the brokers just fine.
> When broker A is restarted, clients fall to the other brokers. Still looks
> good so far.
> If broker B is restarted, the clients do not go back to A they all end up
> on C.
> Having all clients on one broker is not loadbalanced and restarting the
> clents is not an option.
> Did we configure something wrong? would it help to start with different
> sequences in the url (A,B,C), (B,A,C) etc. ?
>
> Regards
>
> Herbert
>
> *Herbert Helmstreit*
> Accenture Industry X | Manufacturing & Operations
>
> Phone: +49 941 / 7 83 92 36
> Herbert.Helmstreit@systema.com
> Accenture GmbH
> Sitz: Kronberg im Taunus. Registergericht: Amtsgericht Königstein im
> Taunus, HRB 5920
> Geschäftsführung: Christina Raab (Vorsitzende der Geschäftsführung), Antje
> Hoffmann, Karl Rathgeb u.a.
> P Please check whether a printout of this e-mail is really necessary.
>
>