Hi,

We're trying to set up a small (3 machine) geoserver cluster, and it's behaving 
oddly - some actions (creating or deleting workspaces, for example) are not 
completed or visible on all nodes in the cluster.

We are using the GeoServer Active Clustering Extension, with an external 
ActiveMQ broker and a shared filesystem between three peers (one of which we 
have set up for management through the UI; the other two are there for HA 
purposes). We are running the cluster in the P2P configuration as suggested 
here: 
https://docs.geoserver.geo-solutions.it/edu/en/clustering/clustering/active/index.html.
 We're running all three nodes as both masters and slaves (as implied by the 
P2P configuration), and they all share the same configuration filesystem. I 
include the cluster configuration settings at the end of this email.

When we create a workspace or a data store, we are seeing occasionally (but 
often enough to be repeatable in a series of manual tests through the REST API) 
that the created resource isn't visible on every node in the cluster. It seems 
like sometimes the synchronisation message doesn't get through to all nodes, 
and they don't pick up the new resource, leading to subsequent failures in 
trying to access it. We can see through the ActiveMQ UI that one of the message 
queues (for one of the workers) isn't receiving the same number of messages as 
the other, which seems suspicious.

The cluster is deployed in Kubernetes on Azure, with each node running in a 
different pod, and the nodes running on ringfenced VMs within the cluster, so 
no other pods will interfere with them. The shared filesystem is a CIFS 
filesystem from Azure, and the ActiveMQ instance is also in its own Kubernetes 
pod (in the same namespace).

We've spent most of the day simply diagnosing this, and we're no further 
forward in working out what's going wrong. Has anybody seen this behaviour 
before, or knows what might be causing it?

Thanks,
Hugo.

Cluster configuration settings:

    CLUSTER_CONFIG_DIR : "/usr/local/geoserver/cluster"
    instanceName : "75f234ba-59db-47e9-937a-d7b44b8b8485"
    readOnly : "disabled"
    empty : "false"
    brokerURL : "tcp://geoserver-broker.geoserver:61616"
    durable : "true"
    embeddedBroker : "disabled"
    toggleMaster : "true"
    connection.retry : "10"
    xbeanURL : "./broker.xml"
    embeddedBrokerProperties : "embedded-broker.properties"
    topicName : "VirtualTopic.geoserver"
    connection : "enabled"
    toggleSlave : "true"
    connection.maxwait : "500"
    group : "geoserver-cluster"



_______________________________________________
Geoserver-users mailing list

Please make sure you read the following two resources before posting to this 
list:
- Earning your support instead of buying it, but Ian Turton: 
http://www.ianturton.com/talks/foss4g.html#/
- The GeoServer user list posting guidelines: 
http://geoserver.org/comm/userlist-guidelines.html

If you want to request a feature or an improvement, also see this: 
https://github.com/geoserver/geoserver/wiki/Successfully-requesting-and-integrating-new-features-and-improvements-in-GeoServer


Geoserver-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-users

Reply via email to