[SFS] Load-balancers, like light-sabers, are (mostly) bullshit

Wed, 10 Jun 2015 14:52:47 -0600 (MDT)

------=_Part_128664_8691737.1433969567455
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit

Yes. No reply at all (h/t Phil Collins) after successful TCP hand-shake (as opposed to denied) is one of the cases where multiple A records is no help, so *don't* tarpit. The other is replies that are valid, but not really valid. 

-- 
David L. Willson 
Teacher, Engineer, Evangelist 
RHCE+Satellite CCAH Network+ A+ Linux+ LPIC-1 UbuntuCP NovellCLA 
Mobile 720-333-LANS(5267) 
http://sofree.us 

This is a good time for a r3VOLution. 

----- Original Message -----

> Wouldn't that depend on some timeout value for the client? What if
> one of the servers hosting www.sofree.us is offline, and connection
> attempts just aren't being responded to as opposed to being actively
> refused? Wouldn't my web browser have to wait for the connection
> attempt to timeout before moving on to the next one? I know web
> developers can be a picky bunch and want pages to load as fast as
> insanely possible because attention spans.

> On Wed, Jun 10, 2015 at 2:26 PM, David L. Willson <
> DLWillson@thegeek.nu > wrote:

> > Go to YouTube and search for "Light-sabers are bullshit"
> 

> > You're welcome. Now, to add to your happy.
> 

> > If you don't do clusters and have a basic understanding of DNS, you
> > probably want to stop reading here.
> 

> > It turns out that having multiple A records for a name does most of
> > what a load-balancer does.
> 

> > Almost every client program (*definitely* web-browsers) are smart
> > enough to try the next address when they're refused connection on a
> > service-port . So, you add two A records with the key "
> > www.sofree.us " and when one web-server stops answering, the
> > *client* will automatically try another address. Yes, really. So,
> > if
> > you have two A records, and http server is down, the clients will
> > automatically move to the other. I always thought that half the
> > clients would just hang or get errror messages. Nope. It just
> > works.
> > No need for haproxy or keepalived or an expensive hardware device,
> > most of the time.
> 

> > Remember that I said that I mostly wanted to work here so I could
> > learn stuph from the brilliant Chris Fedde? I'm paid.
> 

> > --
> 
> > David L. Willson
> 
> > Teacher, Engineer, Evangelist
> 
> > RHCE+Satellite CCAH Network+ A+ Linux+ LPIC-1 UbuntuCP NovellCLA
> 
> > Mobile 720-333-LANS(5267)
> 
> > http://sofree.us
> 

> > This is a good time for a r3VOLution.
> 

------=_Part_128664_8691737.1433969567455
Content-Type: text/html; charset=utf-8
Content-Transfer-Encoding: quoted-printable

<html><head><style type=3D'text/css'>p { margin: 0; }</style></head><body><=
div style=3D'font-family: times new roman,new york,times,serif; font-size: =
12pt; color: #000000'>Yes. No reply at all (h/t Phil Collins) after success=
ful TCP hand-shake (as opposed to denied) is one of the cases where multipl=
e A records is no help, so *don't* tarpit. The other is replies that are va=
lid, but not really valid.<br><br><div><span name=3D"x"></span>--<br>David =
L. Willson<br>Teacher, Engineer, Evangelist<br>RHCE+Satellite CCAH Network+=
 A+ Linux+ LPIC-1 UbuntuCP NovellCLA<br>Mobile 720-333-LANS(5267)<br>http:/=
/sofree.us<br><br>This is a good time for a r3VOLution.<span name=3D"x"></s=
pan><br></div><br><hr id=3D"zwchr"><blockquote style=3D"border-left:2px sol=
id #1010FF;margin-left:5px;padding-left:5px;color:#000;font-weight:normal;f=
ont-style:normal;text-decoration:none;font-family:Helvetica,Arial,sans-seri=
f;font-size:12pt;"><div dir=3D"ltr"><div>Wouldn't that depend on some timeo=
ut value for the client? What if one of the servers hosting <a href=3D"http=
://www.sofree.us" target=3D"_blank">www.sofree.us</a> is offline, and conne=
ction attempts just aren't being responded to as opposed to being actively =
refused? Wouldn't my web browser have to wait for the connection attempt to=
 timeout before moving on to the next one? I know web developers can be a p=
icky bunch and want pages to load as fast as insanely possible because atte=
ntion spans.<br></div></div><div class=3D"gmail_extra"><br><div class=3D"gm=
ail_quote">On Wed, Jun 10, 2015 at 2:26 PM, David L. Willson <span dir=3D"l=
tr">&lt;<a href=3D"mailto:DLWillson@thegeek.nu" target=3D"_blank">DLWillson=
@thegeek.nu</a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote" styl=
e=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div><d=
iv style=3D"font-family:times new roman,new york,times,serif;font-size:12pt=
;color:#000000">Go to YouTube and search for "Light-sabers are bullshit"<br=
><br>You're welcome. Now, to add to your happy.<br><br>If you don't do clus=
ters and have a basic understanding of DNS, you probably want to stop readi=
ng here.<br><br>It turns out that having multiple A records for a name does=
 most of what a load-balancer does.<br><br>Almost every client program (*de=
finitely* web-browsers) are smart enough to try the next address when they'=
re refused connection <b>on a service-port</b>. So, you add two A records w=
ith the key "<a href=3D"http://www.sofree.us" target=3D"_blank">www.sofree.=
us</a>" and when one web-server stops answering, the *client* will automati=
cally try another address. Yes, really. So, if you have two A records, and =
http server is down, the clients will automatically move to the other. I al=
ways thought that half the clients would just hang or get errror messages. =
Nope. It just works. No need for haproxy or keepalived or an expensive hard=
ware device, most of the time.<br><br>Remember that I said that I mostly wa=
nted to work here so I could learn stuph from the brilliant Chris Fedde? I'=
m paid.<br><br><div><span></span>--<br>David L. Willson<br>Teacher, Enginee=
r, Evangelist<br>RHCE+Satellite CCAH Network+ A+ Linux+ LPIC-1 UbuntuCP Nov=
ellCLA<br>Mobile 720-333-LANS(5267)<br><a href=3D"http://sofree.us" target=
=3D"_blank">http://sofree.us</a><br><br>This is a good time for a r3VOLutio=
n.<span></span><br></div><br></div></div></blockquote></div><br></div>
</blockquote><br></div></body></html>
------=_Part_128664_8691737.1433969567455--