Comments on: netstat -s for TCP Statistics for IPv4?

By: bbb

bbb — Sun, 25 Oct 2009 14:12:15 +0000

i thinks its problem ocour, by load on your server

By: Troy Tate

Troy Tate — Tue, 22 Sep 2009 13:09:23 +0000

I would install Microsoft Network Monitor 3.3 and capture traffic during these events. You can analyze it in Network Monitor or use Wireshark to gain additional insight into what is going on. TCP handshakes & RSTs might contribute to some performance issues, but I don’t think the number you mention is excessive when you consider how many packets are actually in a transaction. Are you using large pings (>= 1000bytes)? See if the packet size matters on the network. Maybe there’s a device with bad MTU settings. Also check duplex / speed settings along the path. These should always be hard-set for servers to switch interfaces.

By: tbaror

tbaror — Mon, 21 Sep 2009 12:58:34 +0000

The reason i am asking about Failed Connection Attempts .

trying to identify issue i got with 2003 STD servers.
those servers working with agent that transfer files to remote storage ,
issue appearing each 3~4 days one of the servers 5 of them report fail transfer for almost 1~2 min.
*on the servers itself there is no event that indicate link loos or what
ever for those events
i wrote a script that trying each 1 sec check if path file exist on remote
storage , in case of more than 100ms delay its start fping to see if there
is network delay ,and results form fping are saying everything its ok 0.5
~.0.8 replay.

since i run the script on those servers i have correlated agent events to
script and got about 10 sec and more for file path checking in the same
time.
now when i am doing netstat -es on the servers i get
0 Discards and 0 Errors and only place i get errors is at TCP Statistics for
IPv4
Failed Connection Attempts = 5064
my question is that indicate that the issue exist on the other counterpart
(storage) IO or bandwidth related issue?.
Thanks