RDQM HA architecture

Describes the basic architecture of replicated data queue manager high availability (RDQM HA) configurations to assist with troubleshooting.


Resource names

Various resources are created for each RDQM queue manager and these resources have names based on the Directory name of the queue manager. The name can be found in the file /var/mqm/mqs.ini, and is referred to here as qm. For example, for an RDQM HA queue manager named TMPQM1, qm would be tmpqm1.


Architecture

The architecture of RDQM high availability (HA) involves both DRBD, for data replication, and Pacemaker, for managing where HA RDQM queue managers run.

When creating an RDQM HA queue manager, the following steps are completed:
  1. A DRBD resource is created to replicate the data for the queue manager.
  2. A queue manager is created and configured to use the DRBD resource for its storage.
  3. A set of Pacemaker resources is created to monitor and manage the queue manager.


DRBD

Each RDQM HA queue manager has a DRBD resource file generated for it named /etc/drbd.d/qm.res. For example, when an RDQM HA queue manager named HAQM1 is created, the DRBD resource file is /etc/drbd.d/haqm1.res.

The most important information for troubleshooting purposes in the .res file is the device minor number for this particular DRBD resource. Many of the messages that DRBD logs use this minor number. For the example queue manager, HAQM1, the .res file contains the following information:
device minor 100;
For this queue manager, we should look for messages such as the following example:
Jul 31 00:17:24 mqhavm13 kernel: drbd haqm1/0 drbd100 mqhavm15.gamsworthwilliam.com: drbd_sync_handshake:
The presence of the string drbd100 indicates that the message relates to HAQM1. Not all messages logged by DRBD use the device minor number, some use the DRBD resource name, which is the same as the Directory name of the RDQM HA queue manager. For example:
Jul 31 00:17:22 mqhavm13 kernel: drbd haqm1 mqhavm15.gamsworthwilliam.com: Connection closed


Pacemaker

There are a number of Pacemaker resources generated for an RDQM HA queue manager:

    qm
    This is the main resource representing the RDQM HA queue manager.

    p_rdqmx_qm
    This is an internal resource.

    p_fs_qm
    This is a standard filesystem resource that mounts the volume for the queue manager onto /var/mqm/vols/qm.

    ms_drbd_qm
    This is the master/slave resource for the DRBD resource for the RDQM.

    p_drbd_qm
    This is the primitive resource for the DRBD resource for the RDQM.

If a floating IP address is configured for an HA RDQM then an additional resource is configured:

    p_ip_qm

Parent topic: Troubleshooting RDQM configurations