7064 TCP_KEEPIDLE has needless 10sec minimum

7064 TCP_KEEPIDLE has needless 10sec minimum

[Robert Mustacchi]

https://us-east.manta.joyent.com/rmustacc/public/webrevs/7064/index.html

Work by Patrick Mooney

Re: [developer] 7064 TCP_KEEPIDLE has needless 10sec minimum

[Garrett D'Amore]

[-- Attachment #1: Type: text/plain, Size: 483 bytes --]

Looks ok.  I’d love to hear more about the use case for this.  Keepalives
are used to probe on idle connections.  So if you have no data to send…
normally I see them most useful in keeping NAT gateways from purging a
connection.  10 sec ought to be more than adequate for that use case.

On Mon, Jun 6, 2016 at 10:41 AM, Robert Mustacchi <rm@joyent.com> wrote:

> https://us-east.manta.joyent.com/rmustacc/public/webrevs/7064/index.html
> 
> Work by Patrick Mooney
> 

[-- Attachment #2: Type: text/html, Size: 1044 bytes --]

Re: [developer] 7064 TCP_KEEPIDLE has needless 10sec minimum

[Dan McDonald]

> On Jun 6, 2016, at 3:17 PM, Garrett D'Amore <garrett@damore.org> wrote:
> 
> Looks ok.  I’d love to hear more about the use case for this.  Keepalives are used to probe on idle connections.  So if you have no data to send… normally I see them most useful in keeping NAT gateways from purging a connection.  10 sec ought to be more than adequate for that use case.


IIRC from an older version of this, he's not changing the default, but merely lowering the floor for pathological cases that really need 1sec keepalives.

Dan

Re: [developer] 7064 TCP_KEEPIDLE has needless 10sec minimum

[Patrick Mooney]

Correct.  It was done to keep Linux-y apps happy when they expect and
demand the short interval.  LX is the primary use case, but it
wouldn't surprise me if a few things built natively out of pkgsrc make
the same assumption.

On 6 June 2016 at 14:19, Dan McDonald <danmcd@omniti.com> wrote:
>
>> On Jun 6, 2016, at 3:17 PM, Garrett D'Amore <garrett@damore.org> wrote:
>>
>> Looks ok.  I’d love to hear more about the use case for this.  Keepalives are used to probe on idle connections.  So if you have no data to send… normally I see them most useful in keeping NAT gateways from purging a connection.  10 sec ought to be more than adequate for that use case.
> 
> 
> IIRC from an older version of this, he's not changing the default, but merely lowering the floor for pathological cases that really need 1sec keepalives.
> 
> Dan
> 

Re: [developer] 7064 TCP_KEEPIDLE has needless 10sec minimum

[Josef 'Jeff' Sipek]

LGTM

On Mon, Jun 06, 2016 at 10:41:13 -0700, Robert Mustacchi wrote:
> https://us-east.manta.joyent.com/rmustacc/public/webrevs/7064/index.html
> 
> Work by Patrick Mooney
> 

-- 
Real Programmers consider "what you see is what you get" to be just as bad a
concept in Text Editors as it is in women. No, the Real Programmer wants a
"you asked for it, you got it" text editor -- complicated, cryptic,
powerful, unforgiving, dangerous.

Re: [developer] 7064 TCP_KEEPIDLE has needless 10sec minimum

[Garrett D'Amore]

[-- Attachment #1: Type: text/plain, Size: 1178 bytes --]

I understood all that.  What I mean is, I’m curious about the Linux app
itself that wants this.  I don’t necessarily expect you to know, but if you
do it would be interesting.

On Mon, Jun 6, 2016 at 1:02 PM, Patrick Mooney <patrick.mooney@joyent.com>
wrote:

> Correct.  It was done to keep Linux-y apps happy when they expect and
> demand the short interval.  LX is the primary use case, but it
> wouldn't surprise me if a few things built natively out of pkgsrc make
> the same assumption.
>
> On 6 June 2016 at 14:19, Dan McDonald <danmcd@omniti.com> wrote:
> >
> >> On Jun 6, 2016, at 3:17 PM, Garrett D'Amore <garrett@damore.org> wrote:
> >>
> >> Looks ok.  I’d love to hear more about the use case for this.
> Keepalives are used to probe on idle connections.  So if you have no data
> to send… normally I see them most useful in keeping NAT gateways from
> purging a connection.  10 sec ought to be more than adequate for that use
> case.
> >
> >
> > IIRC from an older version of this, he's not changing the default, but
> merely lowering the floor for pathological cases that really need 1sec
> keepalives.
> >
> > Dan
> >
> 
> 

[-- Attachment #2: Type: text/html, Size: 1881 bytes --]