Summary
DPX jobs, scheduled and run on demand, fail to start and are rescheduled to run 30 minutes later.
Symptoms
In the corresponding data manager log (ssdatmgr.*), the following message is shown:
SNBDBC1008E Error sending message on socket 4, rc=10032
Resolution
DPX maintains a catalog control file snbdb.ini in the bin directory of DPX. This file sets the number and size of memory buffers for each database table. The default value for tables involved in the configuration of DPX, such as the node configuration table ssnode.db, is 20 buffers and 30 KB per buffer. This is a total of 614400 bytes.
In this case, the size of the ssnode table exceeded this maximum allocated memory for the table. The result was excessive processing time that exceeded the timeout value between the data manager and the scheduler, resulting in a rescheduling of the job. This problem can be addressed by increasing the database table parameters through increasing the dbmaxbuff parameter to a value that solves the problem. These parameters are used for all catalog tables. Therefore, you can double or triple this number, but do not exceed the default too much and keep a close eye on server memory used during backup and catalog condense processes.