by Can you help us understand netstat outputs and possibly help
solve a network problem we have?
netstat has a streams allocation section which reports (among
other things) 'total' for each class (size) of packet. What
does this 'total' refer to?
We have been collecting data from netstat trying to analyze a
situation in which the Unix NIC appears to create a spurious
traffic storm (the LED in the concentrator flashes continuously
and collision LEDs flash frequently - no other nodes show
activity) and consumes all the CPU's resources. The system
failed today and we analyzed data for yesterday and today,
taken at 20 minute intervals.
For the class with the highest total figures (class 2 - 64 bytes),
the total rose and fell but the trend was definitely higher. It
peaked at 832,467. For the class with the second highest total
figures (class 7 - 2048 bytes), the total rose or stayed level
- it never fell and peaked at 249,841. If I had just seen the
second pattern, I would have said that the 'total' was cumulative
from Unix re-start. However, the first pattern had peaks and valleys.
Is this a clue to our failure? How do we interpret it? What do
we do about it?
Some other data:
- SCO Unix Sys V on AST Premmia 486/66, 32 MB RAM
- D-Link 32-bit EISA bus mastering Ethernet card serving 15
TCP/IP users on Novell workstations, connecting thru
LANWorkplace (ie: Unix to Novell handles keystrokes and screens
only - low volume)
- the only application is a large (250MB+) Progress database;
Progress also uses the Unix STREAMS facility
- 16 other users access the system through a Computone ISA board
Any help would be greatly appreciated.
----------------------------------------------------------------------
Ottawa, Ontario, Canada 613 825-7608
======================================================================Quote:> Management Systems and Services <
> Specializing in job shops and custom manufacturers. <
> Searching for the perfect software package for the small job shop. <