The IBM WebSphere MQ Recovery Kit uses a number of default values which can be tuned and modified if you have problems with the default settings. The default settings should be sufficient for most environments. If you have problems with timeouts you can use the following table to identify tunable parameters. It is recommended that you do not change the parameters until you have problems with your WebSphere MQ resource hierarchies.
| MQS_QUICKCHECK_TIMEOUT_SC | 10 (seconds) | Timeout for the client connect check. |
| MQS_QUICKCHECK_TIMEOUT_CC | 10 (seconds) | Timeout for the client connect check. |
| MQS_QUICKCHECK_TIMEOUT_PUTGET | 10 (seconds) | Timeout for the PUT/GET check |
| MQS_QUICKCHECK_TIMEOUT_PS | 5 (seconds) | Timeout for the check whether publish/subscribe is in use or not |
| MQS_QUICKCHECK_TIMEOUT_CLUSTER | 5 (seconds) | Timeout for the check whether this queue manager is part of an WebSphere MQ cluster |
| MQS_QUICKCHECK_TIMEOUT | 40 (seconds) | Timeout for the quickCheck script (must be at least 10 seconds). |
| MQS_QMGR_START_TIMEOUT | 60 (seconds) | Timeout for the queue manager start command to complete. |
| MQS_CMDS_START_TIMEOUT | 30 (seconds) | Timeout for the command server start command to complete. |
| MQS_LISTENER_START_TIMEOUT | 30 (seconds) | Timeout for the listener start command to complete |
| MQS_LISTENER_LIST_TIMEOUT | 10 (seconds) | Timeout for the listener list command to complete |
| MQS_CHECK_TIMEOUT_ACTION | ignore | The action in case a server connect check or client connect check times out. The default of “ignore” means that a message about the timeout is logged, but no recovery is initiated. If you set this variable to “sendevent” local recovery is initiated in case a server connect check timed out. |
| MQS_LISTENER_CHECK_DELAY | 2 (seconds) | The time in seconds between the start of the listener and the check for the successful listener start. The default of 2 seconds should be sufficient to detect port in use conditions. |
| NO_AUTO_STORAGE_DEPS | 0 | If you set the variable to 1 the recovery kit does not check if the queue manager and log directory are located in shared storage. If set to 1 the recovery kit does not create file system hierarchies upon resource configuration too. |
| MQS_DSPMQVER_TIMEOUT | 5 (seconds) | Timeout for the dspmqver command (needed to find out the version of WebSphere MQ), must be at least 2 seconds. |
| MQS_SKIP_CRT_MISSING_Q | 0 | Set to 1 to not automatically create a missing test queue. |
| MQS_ALT_USER_NAME | mqm if not set or the user does not have membership in the “mqm” group | The alternate user name to use for all WebSphere MQ commands. By default the user “mqm” is used. If set the alternate user must have its primary group set to the group “mqm” or must have secondary membership in that group. |
To change the parameters add the appropriate variable in the table above to /etc/default/LifeKeeper. The line should have the following syntax:
[…]
MQS_CHECK_TIMEOUT_ACTION=sendevent
[…]
To disable a custom setting and fall back to the default value, just remove the line from /etc/default/LifeKeeper or comment out the corresponding line.
Post your comment on this topic.