[Outages-discussion] Internet Outage Hits Time Warner Customers in southern California - now up

[bobbyjim at gmail.com]

On Aug 29, 2011, at 11:52 AM, "Frank Bulk" <frnkblk at iname.com> wrote:

> Sounds like you can't improve much beyond that!
> 
> Frank
> 
> -----Original Message-----
> From: Dan White [mailto:dwhite at olp.net] 
> Sent: Monday, August 29, 2011 11:38 AM
> To: Frank Bulk
> Cc: 'Jay Ashworth'; outages-discussion
> Subject: Re: Internet Outage Hits Time Warner Customers in southern
> California - now up
> 
> Yes. For equipment that doesn't directly support configuring DHCP relay, we
> can usually position the equipment in our network in such a way that we can
> configure multiple ip helpers on a Cisco switch/router to do the relaying.
> 
> - Dan
> 
> On 29/08/11 11:31 -0500, Frank Bulk wrote:
>> Do you configure your relay points to relay to both DHCP servers, so that
>> you have transparent resiliency?
>> 
>> Frank
>> 
>> -----Original Message-----
>> From: Dan White [mailto:dwhite at olp.net]
>> Sent: Monday, August 29, 2011 11:16 AM
>> To: Frank Bulk
>> Cc: 'Jay Ashworth'; outages-discussion
>> Subject: Re: Internet Outage Hits Time Warner Customers in southern
>> California - now up
>> 
>> Yes. We have a pair of ISC DHCP failover servers that are IPd in different
>> subnets. We use the relay IP of the request to identify which pool to serve
>> the addresses from.
>> 
>> - Dan
>> 
>> On 29/08/11 10:59 -0500, Frank Bulk wrote:
>>> I wonder, can ISC's dhcp failover support operating the DHCP servers in
>>> different subnets?
>>> 
>>> Frank
> 
> 
> _______________________________________________
> Outages-discussion mailing list
> Outages-discussion at outages.org
> https://puck.nether.net/mailman/listinfo/outages-discussion

I used to run @Home.com's. DHCP infrastructure.  We used to get the occasional toilet bowl flush and almost every node would make requests at the same time.  It looked like a DOS and it was a bit of a Microsoft issue in that the client would make a request and if not ack'd it would make another without increasing the timeout.  We used to have to black hole the forwarders and slowly bleed them back.  If you think the customer support  at cable companies is bad,  try partnering with them.  

Rob