|
HP OpenVMS Systemsask the wizard |
|
The Question is: Hello Wizard, I wonder if you can help me with this problem. I have a cluster with includes three boot nodes. Two of them are Alpha's running OpenVMS AXP 6.2 and one which is the one giving me problems is a uVAX II running VMS 5-5.2. The uVAX is giving me PEA0: erro rs which creep up slowly. I am now at 30 errors since by last reboot four days ago. I must say that a lot of data is being transferred from dual ported disks which the uVAX has with a PDP and the Alpha. Another indication of my problems is when I do a "show line qna-0 counter" in NCP. the "buffer unavailable counter" is around 15,000 now. I thought that I may have a defective ethernet card and I replaced it. That did not make a difference. Also, the Alpha are not showing wny of these networking er rors. I have contacted Dec Support and they say that I should not have these buffer anavailable errors, however they do say that the system is a slower one and should probably be replaced. Any ideas? Hope you can help, thank you. The Answer is : There are two DECnet "buffer unavailable" errors: The first, "system buffer unavailable", is typically caused by more data than the system can handle -- DECnet is overrunning the device driver data buffers. The typical fixes are to reduce the traffic or to configure faster CPUs or to unload the existing CPUs of other activity, assuming that basic system and application tuning has been performed. The second, "user buffer unavailable", typically indicates that the application itself cannot keep up with the traffic. Potential solutions to this problem typically involve removing the bottleneck -- this could include tuning or reworking the application, faster CPUs, or other changes.
|