Task #9368
closed
JackRabbit Repo cluster over Postgres for development infra
Added by Massimiliano Assante almost 8 years ago.
Updated over 7 years ago.
Assignee:
_InfraScience Systems Engineer
Infrastructure:
Development
Description
The current JackRabbit Repository instance for development is going to be dismissed and will be replaced by the same setting we have in production. With the only difference that the cluster is going to be fully operative.
We have 2 nodes already available:
workspace-repository1-d.d4science.org
workspace-repository2-d.d4science.org
And we are going to use postgresql-srv-dev.d4science.org for the DB.
- Description updated (diff)
- Infrastructure deleted (
Pre-Production)
I need the correct scopes
Another piece: the version I have in the playbook is 1.8.0-4.6.0-149356 but I'm sure that a newer version was manually installed. What's the latest good one?
- Status changed from New to In Progress
The hostname configured on the load balancer is workspace-repository.dev.d4science.org.
As far as the list of scopes
/gcube
/gcube/devsec
/gcube/devNext
but calls coming from vres must be accepted as well (authorizedChildren property set to true in container.xml)
- Status changed from In Progress to Feedback
- % Done changed from 0 to 90
Both servers have been provisioned.
The paths to the jackrabbit data are identical to the production one, with /home/gcube/tomcat/jacrabbit that points to /data/jackrabbit (no dedicated file system, this time).
I stopped tomcat on workspace-repository2-d.d4science.org (workspace-repository-devsec.d4science.org) so that there's a chance to test the cluster setup.
About the cluster: I don't know if the cluster ID must be numeric: I set it to the VM hostname.
I also already set the janitorEnabled value to true on both servers.
Remember to not start the second server before adding the required cluster information into the database.
- % Done changed from 90 to 0
Andrea Dell'Amico wrote:
Both servers have been provisioned.
The paths to the jackrabbit data are identical to the production one, with /home/gcube/tomcat/jacrabbit that points to /data/jackrabbit (no dedicated file system, this time).
I stopped tomcat on workspace-repository2-d.d4science.org (workspace-repository-devsec.d4science.org) so that there's a chance to test the cluster setup.
About the cluster: I don't know if the cluster ID must be numeric: I set it to the VM hostname.
I also already set the janitorEnabled value to true on both servers.
Remember to not start the second server before adding the required cluster information into the database.
Thanks Andrea, we will report here the results after some tests
- % Done changed from 0 to 90
Yesterday we take some hours to add the second node to the cluster and start up it as well, make some tests and so on. The configuration seems ok, and there was no problem in adding it despite the fact that the cleaner thread was active since the beginning. Another test I did was to check how the cleaner behaves when one of the node is down (but still belongs to the cluster): it just cleans up the revision information no longer needed by both nodes (that's reasonable).
However, I've removed again workspace-repository-devsec.d4science.org from the cluster, since we would like to have the storage cleaned before we go further with tests.
P.S.
to remove a node from the cluster, its node_id must be removed from local_revisions table
The infrastructure is in place since months. The cluster features aren't used yet but it's not a matter of configuration.
- Status changed from Feedback to Closed
- % Done changed from 90 to 100
Also available in: Atom
PDF
Updated by 
Updated by 
Updated by Costantino Perciante 
