How to resolve the "Fail" in Netstat -m

[salimwng]

Hello,

After using the command netstat -m on the command prompt of unix ( SCO Unix Sys V - Rel 3.2.4.2 ) i obtain many failures in some of classes together with the dblks. How can i prevent that ? The problem on our system is that intermitently we have 4 to 9 users that are kicked out all of a sudden. The 4 to 9 users are not the same people, hence it's not a problem of their NIC card. In total we have around 87 users working concurrently among which are those who updates and those who are normal viewers only. We have been having this problem since quite a while now and after many search made, thought i even change the NIC of the Unix box without success, i fall on the netstream.

The users who are kicked out, do get the Unix login few seconds after. When sometimes they don't, i have to stop tcpip manually from the command prompt and re-start it again by using ;

root> tcp stop
root> tcp start

I have posted the netstat -m output below, can anybody help me in figuring out what could be the problem ?

root> netstat -m

streams allocation:
config alloc free total max fail
streams 160 41 119 5394 96 0
queues 864 198 666 16105 592 0
mblks 1855 128 172719576210 801 0
dblks 1484 128 135614476706 698 41152
class 0, 4 by 296 29 267 264690 237 30212
class 1, 16 by 296 9 287 232829 236 70
class 2, 64 by 296 16 28012908609 254 440
class 3, 128 by 264 57 207 607481 226 358
class 4, 256 by 136 17 119 87105 122 589
class 5, 512 by 72 0 72 14165 57 0
class 6,1024 by 52 0 52 4004 22 0
class 7,2048 by 60 0 60 39836 49 0
class 8,4096 by 12 0 12 317987 9 9483

total configured streams memory: 452.45KB
streams memory in use: 23.67KB
maximum streams memory used: 319.11KB

Thank you.
Salim

 

[PHV]

You have to increase the stream resources by tuning the following kernel parameters:
The NSTREAM value will impact the Class* values.
You have to increase the NSTRPAGES value too

Hope This Help, PH.
Want to get great answers to your Tek-Tips questions? Have a look at FAQ219-2884