Original Message:
Sent: Wed August 14, 2024 08:24 AM
From: Martin Evans
Subject: What type of storage are you using for MQ Native HA on Openshift/k8s?
Hi Andres,
I think the use of "block" storage can sometimes get a bit misunderstood in OpenShift and Kubernetes, Native HA uses block storage with a filesystem, and only RWO is required, meaning the disk is only accessible by one host at a time. I think sometimes people confuse "block" with raw-block i.e. no filesystem because the application is determining how data is laid down on the disk, IBM MQ does not do that. The term Files or File storage seems to be used to describe something like NFS shared storage, and RWX is used to select or describe that. So in your case you have selected block storage that will have a filesystem like XFS or EXT4 on it and will only be mounted to a single host (RWO). Hope that helps.
------------------------------
Regards,
Martin Evans
IBM MQ Technical Product Manager
Original Message:
Sent: Wed August 14, 2024 06:09 AM
From: Andres Colodrero
Subject: What type of storage are you using for MQ Native HA on Openshift/k8s?
Hi Martín,
In my original message i was planning to use MQ Native HA in a cluster, with DR (replicating volumes) to a passive to a secundary native MQ. But as IBM MQ has plans to replicate data from one cluster to another (in separate OPenshift clusters), im not longer planning for DR with ODF:
On my tests im using OPenshift on VmWare. The storage class support both filesystem and block.
storage:
defaultClass: thin-csi
persistedData:
deleteClaim: true
enabled: false
queueManager:
class: thin-csi
size: 20Gi
type: persistent-claim
The operator is creating PVCs type: RWO and Filesystem.
I wonder is this is correct once the documentation mention Block Storage provider
------------------------------
Andres Colodrero
Original Message:
Sent: Tue August 13, 2024 05:16 AM
From: Martin Evans
Subject: What type of storage are you using for MQ Native HA on Openshift/k8s?
Hi Andres,
One thing to consider with ODF is that you might be replicating the replication that IBM MQ Native HA is performing. As IBM MQ is already replicating your data you could look at using a simpler non-software-defined storage (SDS) solution. In my test OpenShift cluster I use the LVM Operator and locally attached storage which uses a lot less resource than running ODF.
------------------------------
Regards,
Martin Evans
IBM MQ Technical Product Manager
Original Message:
Sent: Thu June 27, 2024 05:04 AM
From: Andres Colodrero
Subject: What type of storage are you using for MQ Native HA on Openshift/k8s?
Hi,
Documentation mention that RWO volumes should be used with those conditions:
Obtained from a block storage provider.
Formatted as ext4 or XFS.
Supports dynamic volume provisioning
what are you using to provision this? I have tested ODF, but for running file/block storage requires a lot of resources.
------------------------------
Andres Colodrero
------------------------------