03-25-2019 12:08 PM
Hello,
I use :
I plan to deploy a ** HA Nuxeo Cluster**.
If i use org.nuxeo.ecm.core.storage.sql.kv.SQLKeyValueStore
according to https://jira.nuxeo.com/browse/NXP-25604 , do i need Redis ?
What i understood (reading the documentation), if i use "SQLKeyValueStore" i didn't need Redis, but i have to use Kafka. Is it right ?
Thank you for your help.
03-25-2019 05:54 PM
Indeed this page is not up to date (we'll update it) and as you correctly found out SQLKeyValueStore
can be used for the key/value store. And actually it is automatically used if you don't specify nuxeo.keyvalue.provider=mongodb
(done by default by the mongodb
template) or nuxeo.keyvalue.provider=redis
(done by default by the redis
template).
Note that in a cluster setting you also need a pub/sub provider, which can be either Redis or Stream. If you use the redis
template then the default is nuxeo.pubsub.provider=redis
, otherwise if you specify nuxeo.pubsub.provider=stream
a Stream-based implementation will be used; specify kafka.enabled=true
for the Kafka implementation otherwise the default will be Chronicle Queue which is not cluster-aware (streams aren't shared between nodes). If you don't specify any nuxeo.pubsub.provider
then you'll get a memory-based one which isn't cluster-aware either, nor is it resilient to server restarts.
03-25-2019 01:02 PM
03-25-2019 01:52 PM
Thank you for the link, but i don't use MongoDB but only Postgresql.
03-25-2019 05:54 PM
Indeed this page is not up to date (we'll update it) and as you correctly found out SQLKeyValueStore
can be used for the key/value store. And actually it is automatically used if you don't specify nuxeo.keyvalue.provider=mongodb
(done by default by the mongodb
template) or nuxeo.keyvalue.provider=redis
(done by default by the redis
template).
Note that in a cluster setting you also need a pub/sub provider, which can be either Redis or Stream. If you use the redis
template then the default is nuxeo.pubsub.provider=redis
, otherwise if you specify nuxeo.pubsub.provider=stream
a Stream-based implementation will be used; specify kafka.enabled=true
for the Kafka implementation otherwise the default will be Chronicle Queue which is not cluster-aware (streams aren't shared between nodes). If you don't specify any nuxeo.pubsub.provider
then you'll get a memory-based one which isn't cluster-aware either, nor is it resilient to server restarts.
03-26-2019 05:14 AM
Thank you for this detailled answer, It's very clear !
03-26-2019 05:53 AM
We recommend Kafka. Redis cannot do everything related to Nuxeo Stream features.
03-26-2019 09:56 AM
Thank you ! so I will follow your recommendations and go with Kafka !
Find what you came for
We want to make your experience in Hyland Connect as valuable as possible, so we put together some helpful links.