[Info-vax] TCPIP RWAST

[Jeffrey H. Coffield]

On 03/09/2021 02:46 PM, Dave Froble wrote:
> On 3/9/2021 4:20 PM, Jeffrey H. Coffield wrote:
>>
>>
>> On 03/09/2021 12:05 PM, Stephen Hoffman wrote:
>>> On 2021-03-09 19:48:05 +0000, Jeffrey H. Coffield said:
>>>
>>>>  FIN_WAIT_2
>>>
>>> Some number of connections in that state is to be expected, pending
>>> timeout.
>>>
>>> App crashes or network disconnections or misconfigured firewalls can or
>>> will exacerbate those numbers.
>>>
>>> https://benohead.com/blog/2013/07/21/tcp-about-fin_wait_2-time_wait-and-close_wait/
>>>
>>>
>>>
>>>
>>>
>> This may be the answer, I checked all the other vms systems I have
>> access to and the number of FIN_WAIT_2 connections is 0 to 2. The system
>> in question has 975.
>>
>> I have a conference call with with the network people today. I'll update
>> this thread with any results.
>>
>> Many thanks
>> Jeff
>
> Might be something there, especially with the history.
>
> I believe your wait is what TCP/IP does before taking down the device.
> Noise on the network, not sure what that might be, could be keeping the
> devices allocated.
>
> We have had some issues with the timeout in the past.  Might have
> changed parameters to make the timeout shorter.
>
> I'm wondering if a duplex mismatch could cause such problems?
>

When the system was moved and when the switches were upgraded I made 
sure both the system and the switch showed 1000Mb/sec full duplex.

The link Stephen Hoffman gave steered me in this direction. A trouble 
ticket has been raised with the firewall group to see if they are the 
culprit.

Jeff