"unable to alloc 85588469 bytes" message

[John Perks]

Hello,

I have an outbound thread which gets the message “unable to alloc nnnnnnnn bytes”, where nnnnnnnn is usually > 64Mg.

This interface is running on AIX 5.2, Quovadx 5.2.1P2.

The outbound interface protocol is MQ using the server on the same system.

A TPS Outbound Data tcl proc is uuencoding jpg files, putting them into OBX.5; hence the rather large outbound messages.

Typically this error occurs at about 85Mg, but it has occured at as low as 34Mg in the past.

This interface is the only thread in the process.

MQ is configured to handle up to 100Mg, and the procedure which creates the OBX with the uuencoded images is hard-coded to kill the message if the resultant message is over 100Mg.  The message is not getting as far as MQ; by trial and error and tcl proc logging it has been determined that the “unable to alloc” message occurs as the tcl proc is creating or continuing the message.

When the error occurs we have had to skip the message and restart the interface, effecting only this process and thread.  But more recently, upon restart, we get a 921 DbVista error and resort to re-initializing the database.

Any advice or similar past experiences that could lead to a fix or workaround would be appreciated.

Thanks.

Topic