Retries are controlled by WTRETRIES and MAXRETRIES on Connect:Direct for z/OS. Here is what you should know.

Body

When you are a C:D z/OS PNODE initiating a session to another node, retries will be attempted for a failed connection.

These Init Parms control how often and how many retries will occur.

WTRETRIES = hh:mm:ss – This is the time between retries (Default = 3:00). Processes in the timer retry queue (TI RE) will be attempted at the next WTRETRIES interval.

MAXRETRIES = number of retries – This is the maximum number of times that a retry will be attempted. This is a number in the range of 0-512 (Default = 7). You can code 0 here to disable automatic retries.

What happens when retries (MAXRETRIES) are exhausted?

When all retries have been attempted, the process goes into the hold waiting connection queue (HO WC) with a SVTM105I where it will not be reattempted until a successful connection is made or manual intervention occurs.  

What happens when there are multiple failed processes to the same SNODE?

If there are multiple processes for the same node, only the first will retry. The others go into HO RH for Restart Held. They will start once another process for that node successfully starts. This is called Intelligent Session Retry. It prevents multiple processes going to the same node from retrying until the connection is reestablished.

Once the PROCESS is in HO WC, any of the following things will cause it to attempt execution:

1. The process is manually held, then released.

2. An incoming connection is established by the remote node.

3. A different process submitted on this node successfully establishes an outbound connection to the remote node.

4, The Connect:Direct DTF is restarted.

Example:

WTRETRIES = 5:00

MAXRETRIES = 3

The first failure will go into a 5 minute wait on TI RE for the WTRETRIES period. If that retry fails, another 5 minutes elapses before the next retry. This will occur for up to 3 cycles, after which the process goes into the HO WC. If there are multiple failed processes for that SNODE, only the first is retried. The remaining processes go to HO WH until the connection is reestablished.