Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 20 Jan 2013 20:12:09 -0600
From:      Adam Vande More <amvandemore@gmail.com>
To:        Karthik Reddy <22karthikreddy@gmail.com>
Cc:        freebsd-questions@freebsd.org
Subject:   Re: What is the timeout of TCP in freeBSD?
Message-ID:  <CA%2BtpaK3mP06wt=3G5GYN8EV1pvYsn2Eq070Fuo-eHq_4UJExjw@mail.gmail.com>
In-Reply-To: <CALK7g0h82SJSObmVo08z6=UAgB-ykc3QS4ApCAnkcvxbCdy5UA@mail.gmail.com>
References:  <CALK7g0h82SJSObmVo08z6=UAgB-ykc3QS4ApCAnkcvxbCdy5UA@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, Jan 20, 2013 at 8:50 AM, Karthik Reddy <22karthikreddy@gmail.com>wrote:

> I was doing a experiment on FreeBSD for testing TCP timeout and RTO. OS is
> being run from two different VMware versions 4.0 and 5.0.
>
> Present Scenario: VMware Player 4.0
> I'll start a telnet session to a non-existing system in the network. When I
> look at the tcpdump the RTO starts at every 3 seconds and after some
> exponential backoff starts. In this scenario after 75 seconds the TCP gives
> up and tells me that there is no system existing with the IP and telnet
> session terminates.
>
> Next Scenario: VMware Player 5.0
> In this scenario, I did the same but the RTO starts at 5 sec and then
> varies. In this scenario, it takes more than 120 seconds for telnet session
> to tell me that there is no system is available in the network.
>
> I have seen sysctl in both VM's. net.inet.tcp.keepinit = 75000
>
> Is this problem something related to timing of the VM's or any other issue?
>

What's the wallclock delta during such a test?  Have you tried setting
'kern.hz="50"' or fiddling other TC options?  UP VM's tend to keep time
better than other multicore configs.

-- 
Adam Vande More



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CA%2BtpaK3mP06wt=3G5GYN8EV1pvYsn2Eq070Fuo-eHq_4UJExjw>