Hello
I have a 3.2.4.2 SCO UNIX server that has in the past recieved thousands of failures in the netstat -m fail column. This customer has this server on a VLAN but still had been expierienced the problem. My expierience is that this is related to bradcast messages and information recieved from the network that causes this issue. I manually modified the streams memory to allow more time for the CPU to parse the data being recieved from the network. It seems that I have modified it enough now that the server is stable. However should I really ever have to do this. The customer is now pointing blame at the UNIX server as having the problem, but I maintain is is caused by network issues and I have just patched their network issue by allowing more time to parse the data. Am I correct in my assesment of the problem and have any of you ever seen this type of failure before? I understand that Openserver 5 is much more auto tuning in the streams memory area and would probably handle this failure better but I need a second opinion to ease my assesment of the problem.
Any respond would be appreciated and Thanks in advance.
I have a 3.2.4.2 SCO UNIX server that has in the past recieved thousands of failures in the netstat -m fail column. This customer has this server on a VLAN but still had been expierienced the problem. My expierience is that this is related to bradcast messages and information recieved from the network that causes this issue. I manually modified the streams memory to allow more time for the CPU to parse the data being recieved from the network. It seems that I have modified it enough now that the server is stable. However should I really ever have to do this. The customer is now pointing blame at the UNIX server as having the problem, but I maintain is is caused by network issues and I have just patched their network issue by allowing more time to parse the data. Am I correct in my assesment of the problem and have any of you ever seen this type of failure before? I understand that Openserver 5 is much more auto tuning in the streams memory area and would probably handle this failure better but I need a second opinion to ease my assesment of the problem.
Any respond would be appreciated and Thanks in advance.