HP OpenVMS Systemsask the wizard |
The Question is: Every now and then my system goes into a deadlock because og Nonpaged Dynamic Memory leak. Something eats up the pool very fast expecially when no one is using it f.ex. over a weekend and when we touch it on monday morning all gets stuck and we have to re boot. In the error log we always get error message from the KFMSA sub system indecating Virtual Circuit Timeout prior to the stop. We have two VAX-6000 on double lan but no cluster. The KFMSA is part of a DSSI to SCSI where four optical disks ar connected to both the systems. I wonder if you have a hint for me about this problem. The Answer is : A pool leak implies a problem in privileged-mode (executive or kernel) software -- this could be a bug in OpenVMS itself, or this could be a bug in a layered product, or in local or a third-party software package containing (buggy) privileged-mode code. Quite obviously, you will have to find and plug the non-paged pool leak. One obvious approach would be to (try to) determine what specific data structure(s) are filling up the non-paged pool, and back-track this to the culprit package. You will also want to consider what has changed prior to this problem arising -- what ECO kits or upgrades or new packages have been installed; what hardware or software changes might have precipitated this leak. You will likely also want to upgrade to a rather more current OpenVMS VAX version, and to more current (privileged) layered product versions, and apply the current ECO kits for the versions. V6.1 is rather old and is no longer supported, and an upgrade to V7.2 is recommended and is expected to have no particular impact on user-mode or privileged-mode OpenVMS VAX applications.
|