Re: [AFMUG] Cost of lashing onto strand

[Steve Jones]

doesnt sound unreasonable for a stand alone service. Doesnt seem too
complicated for a company with the truck, the lasher, the labor, the
experience and the insurance. The amigos are probably 25-50 cents a foot.
Johnny with a hook ladder and 30 pack is probably 75 cents.

A lot on that estimate depends on the caliber of the contractor

On Tue, Feb 11, 2025 at 6:09 PM Adam Moffett  wrote:

> Well there is a little more to it than just pulling the rope.  In any
> case, the prices on this guy's spreadsheet make me think engineering was a
> bad career choice and I should have invested in a bucket truck.
>
> --
> *From:* AF  on behalf of Dev <
> d...@logicalwebhost.com>
> *Sent:* Tuesday, February 11, 2025 6:59 PM
> *To:* AnimalFarm Microwave Users Group 
> *Subject:* Re: [AFMUG] Cost of lashing onto strand
>
> Seems like a ripoff, all you do is walk down the road pulling a thingy
> with ropes that does the twirling of wire while you walk, seems like the
> cost to walk in a straight line should be cheap. Maybe they’re walking with
> golden boots?
>
> On Feb 11, 2025, at 3:50 PM, Adam Moffett  wrote:
>
> I haven't had a quote from a contractor in front of me for about 10 years,
> but I remember the lashing was $0.25/foot.
>
> I just saw someone's cost analysis spreadsheet showing $1.30/ft for
> lashing.  The actual cable is separate.  The strand installation is
> separate.  That's $1.30/ft for just lashing.  Has it really gone up 5x in
> 10 years or is this guy getting screwed?
>
> -Adam
>
> --
> AF mailing list
> AF@af.afmug.com
> http://af.afmug.com/mailman/listinfo/af_af.afmug.com
>
>
> --
> AF mailing list
> AF@af.afmug.com
> http://af.afmug.com/mailman/listinfo/af_af.afmug.com
>
-- 
AF mailing list
AF@af.afmug.com
http://af.afmug.com/mailman/listinfo/af_af.afmug.com

Re: [AFMUG] DHCP Fail over

[Dennis Burgess - LTI Support via AF]

You can also run a mikrotik DHCP Server with relay going to each server, the MT 
server can run virtually and have high availability on itself, but the three 
DHCP (relays), will all be pulling from the same pool.

From: AF  On Behalf Of Josh Luthman
Sent: Tuesday, February 11, 2025 6:09 PM
To: AnimalFarm Microwave Users Group 
Subject: Re: [AFMUG] DHCP Fail over

Kea is what you want, I think...

https://www.isc.org/kea/

For HA: 
https://kea.readthedocs.io/en/latest/arm/hooks.html#supported-configurations

On Tue, Feb 11, 2025 at 6:23 PM Adam Moffett 
mailto:dmmoff...@gmail.com>> wrote:
We have two DHCP servers per market and they run VRRP.  VRRP gives you an 
active/standby setup.  Configurations have to be synchronized of course, but 
I'd say this is the simplest way.

To have any kind of active/active setup the DHCP servers would have to share 
the same lease database.  I believe ISC had a way to do that where they would 
send messages to update each other, but I haven't looked into this in awhile so 
I may be hallucinating that.

-Adam



From: AF mailto:af-boun...@af.afmug.com>> on behalf of 
Jesse DuPont 
mailto:jesse.dup...@celeritycorp.net>>
Sent: Tuesday, February 11, 2025 5:29 PM
To: AnimalFarm Microwave Users Group mailto:af@af.afmug.com>>
Subject: Re: [AFMUG] DHCP Fail over

What will you be using for your DHCP "concentrator" (for lack of a better 
term); that is, what will be the gateway device(s)? It seems you'd be better 
served by having a pair of routers running VRRP or some other cluster prototol, 
then having redundant DHCP servers that the concentrators/gateways relay to 
simultaneously (both of which check with RADIUS for auth and assignment for 
statics). The two DHCP servers can be configured active/active or active/backup 
and they'll both serve the same blocks (based on what RADIUS tells them to 
provide). ISC DHCP did this "okay", but KEA DHCP (ISC's replacement) does it 
really well. The two gateways using VRRP would appear like a single device and 
have a single IP. Depending on the routers, sometimes "state" (like current ARP 
resolutions) are sync'd between both routers, sometimes the failover router has 
to just re-ARP for everything; not the end of the world.

You can simplify all this by using an actual BNG for your DHCP side (and your 
PPPoE, for that matter). Something like NetElastic's or IP Infusion's BNG can 
do all this.
On 2/11/25 3:12 PM, Mark - Myakka Technologies via AF wrote:

We currently run 3 PPPoE servers using an OSPF concentrator and radius to 
manage the IP addresses.  With this setup, it doesn't matter which IP lands on 
which PPPoE server.  OSFP handles it.



We now need to do something similar with DHCP.  I've been messing around with 
/32's and Option 121, but just can not get a stable solution.  I'm now thinking 
about plan B.  Similar general setup we use on the PPPoE side.  Lets say we go 
with 3 DHCP servers connected to an OSPF concentrator.  I would have to set my 
DHCP network on all 3 servers to something like 
192.168.0.0/23 for about 512 address total.  Server one 
will do a GW of 192.168.0.1, Server two will do a GW of 192.168.0.2, server 3 
will do a GW of 192.168.0.3.  When a client connects they will randomly connect 
to one of the 3 servers and receive an IP address from radius.  My current 
thoughts are



1. Each server will have a /32 address not the /23.  IP address on server 1 
will be 192.168.0.1/32.

2. OSFP will only announce the /32 address of the server to the concentrator.

3. I will have to use the DHCP script option to insert and delete the clients 
ip address as a /32 in OSPF on the server to update the concentrator.



The one issue I see off the bat is when a client reboots.  If the client 
reboots and moves from server 1 to server 3, I now have two servers with the 
same IP address.  I think I can deal with that by using a short lease time.





Thoughts?  I'm still digging around looking for other (better) options of 
having DHCP fail-over.  The one option that will not work is reserving a block 
of IPs per server.  We have several customers that are using static IPs, so 
they need to be accessible from all 3 servers.





--



Thanks,

 Mark  mailto:m...@mailmt.com



Myakka Communications

https://imsva91-ctp.trendmicro.com:443/wis/clicktime/v1/query?url=www.Myakka.com&umid=8BF5F074-2DF8-1D06-9B77-963F7B157DC1&auth=079c058f437b7c6303d36c6513e5e8848d0c5ac4-69f5912742fe569d6e9a6efd4038ce9390b19830



Serving Manatee and Sarasota Counties with High-Speed Internet for over 20 years





--
AF mailing list
AF@af.afmug.com
http://af.afmug.com/mailman/listinfo/af_af.afmug.com
-

Re: [AFMUG] DHCP Fail over

[Dennis Burgess - LTI Support via AF]

One downside to going DHCP vs PPPoE.  If bandwidth is under 400 meg, PPPoE is 
the way to go.

From: AF  On Behalf Of Sterling Jacobson via AF
Sent: Wednesday, February 12, 2025 3:42 PM
To: AnimalFarm Microwave Users Group 
Cc: Sterling Jacobson 
Subject: Re: [AFMUG] DHCP Fail over

Dennis, isn't that a recipe for double IP assignments?

Wouldn't each DHCP server (relay DHCP endpoint server) need to have 
non-overlapping IPv4 pools?

ASFAIK there is no actual HA replication of DHCP tables on a server, so if one 
server is always responding to the layer2 domain request and it becomes 
unavailable the secondary or tertiary server would answer with a stale table 
and possibly assign a duplicate?


From: AF mailto:af-boun...@af.afmug.com>> on behalf of 
Dennis Burgess - LTI Support via AF mailto:af@af.afmug.com>>
Sent: Wednesday, February 12, 2025 1:52 PM
To: AnimalFarm Microwave Users Group mailto:af@af.afmug.com>>
Cc: Dennis Burgess - LTI Support 
mailto:dmburg...@linktechs.net>>
Subject: Re: [AFMUG] DHCP Fail over


You can also run a mikrotik DHCP Server with relay going to each server, the MT 
server can run virtually and have high availability on itself, but the three 
DHCP (relays), will all be pulling from the same pool.



From: AF mailto:af-boun...@af.afmug.com>> On Behalf Of 
Josh Luthman
Sent: Tuesday, February 11, 2025 6:09 PM
To: AnimalFarm Microwave Users Group mailto:af@af.afmug.com>>
Subject: Re: [AFMUG] DHCP Fail over



Kea is what you want, I think...



https://www.isc.org/kea/



For HA: 
https://kea.readthedocs.io/en/latest/arm/hooks.html#supported-configurations



On Tue, Feb 11, 2025 at 6:23 PM Adam Moffett 
mailto:dmmoff...@gmail.com>> wrote:

We have two DHCP servers per market and they run VRRP.  VRRP gives you an 
active/standby setup.  Configurations have to be synchronized of course, but 
I'd say this is the simplest way.



To have any kind of active/active setup the DHCP servers would have to share 
the same lease database.  I believe ISC had a way to do that where they would 
send messages to update each other, but I haven't looked into this in awhile so 
I may be hallucinating that.



-Adam







From: AF mailto:af-boun...@af.afmug.com>> on behalf of 
Jesse DuPont 
mailto:jesse.dup...@celeritycorp.net>>
Sent: Tuesday, February 11, 2025 5:29 PM
To: AnimalFarm Microwave Users Group mailto:af@af.afmug.com>>
Subject: Re: [AFMUG] DHCP Fail over



What will you be using for your DHCP "concentrator" (for lack of a better 
term); that is, what will be the gateway device(s)? It seems you'd be better 
served by having a pair of routers running VRRP or some other cluster prototol, 
then having redundant DHCP servers that the concentrators/gateways relay to 
simultaneously (both of which check with RADIUS for auth and assignment for 
statics). The two DHCP servers can be configured active/active or active/backup 
and they'll both serve the same blocks (based on what RADIUS tells them to 
provide). ISC DHCP did this "okay", but KEA DHCP (ISC's replacement) does it 
really well. The two gateways using VRRP would appear like a single device and 
have a single IP. Depending on the routers, sometimes "state" (like current ARP 
resolutions) are sync'd between both routers, sometimes the failover router has 
to just re-ARP for everything; not the end of the world.

You can simplify all this by using an actual BNG for your DHCP side (and your 
PPPoE, for that matter). Something like NetElastic's or IP Infusion's BNG can 
do all this.

On 2/11/25 3:12 PM, Mark - Myakka Technologies via AF wrote:

We currently run 3 PPPoE servers using an OSPF concentrator and radius to 
manage the IP addresses.  With this setup, it doesn't matter which IP lands on 
which PPPoE server.  OSFP handles it.



We now need to do something similar with DHCP.  I've been messing around with 
/32's and Option 121, but just can not get a stable solution.  I'm now thinking 
about plan B.  Similar general setup we use on the PPPoE side.  Lets say we go 
with 3 DHCP servers connected to an OSPF concentrator.  I would have to set my 
DHCP network on all 3 servers to something like 
192.168.0.0/23 for about 512 address total.  Server one 
will do a GW of 192.168.0.1, Server two will do a GW of 192.168.0.2, server 3 
will do a GW of 192.168.0.3.  When a client connects they will randomly connect 
to one of the 3 servers and receive an IP address from radius.  My current 
thoughts are



1. Each server will have a /32 address not the /23.  IP address on server 1 
will be 192.168.0.1/32.

2. OSFP will only announce the /32 address of the server to the concentrator.

3. I will have to use the DHCP script option to insert and delete the clients 
ip address as a /32 in OSPF on the server to update the concentrator.



The one issue I see off the bat is when a client reboots.  If the client 
reboots and moves from server 1 

Re: [AFMUG] hired manager

[Ken Hohhof]

Maybe management has changed since I was doing it, also depends on whether
you are first level, middle manager, C-suite, line management or staff, etc.
But I had a pretty full day.

 

Hiring, firing, quarterly performance reviews, salary increases, promotions,
assigning projects

Project management

Budgets (capital, salaries, expenses)

Attending project reviews, design reviews, staff meetings

Writing monthly project and budget status reports for upper management

Approving things like purchase orders, vacation requests, etc.

Try to offload meetings and paperwork from your people so they can do actual
work

 

Much of the time was problem solving related to project or personnel issues.
Prototype fails testing, need an unexpected PCB layout cycle, how to keep
project from slipping and affecting overall program.  Or a key person quits
or gets injured or an employee can't handle an assignment, do you assign a
more senior person to help, hire a contractor, shuffle assignments, or just
accept a slip in the schedule, etc.  Then there's managing your manager.
Example from real life - senior management says cut 10% of employees but
decides not to do it my seniority or skill level but by canceling projects
and then telling us to fire whoever was on those projects.  Except that has
us keeping an employee we planned to fire for poor performance and firing
the senior person we had assigned to help the incompetent one, dooming both
projects, so we need to convince the big boss to let us decide who to fire
and who to keep.  My impression is big tech companies no longer worry about
this, they have some AI program pick 1,000 random employees and email or
text them they're fired.

 

From: AF  On Behalf Of Sterling Jacobson via AF
Sent: Wednesday, February 12, 2025 4:27 PM
To: AnimalFarm Microwave Users Group 
Cc: Sterling Jacobson 
Subject: Re: [AFMUG] hired manager

 

Ummm, I'm going to need you to come in on Sunday...

 

Steve, you are right about your observations from my limited experience.

That is why fractional is becoming more of an option, and it should be.

 

It's also why management needs more bodies to fill their time managing them
lol

 

  _  

From: AF mailto:af-boun...@af.afmug.com> > on
behalf of Ken Hohhof mailto:khoh...@kwom.com> >
Sent: Wednesday, February 12, 2025 3:05 PM
To: 'AnimalFarm Microwave Users Group' mailto:af@af.afmug.com> >
Subject: Re: [AFMUG] hired manager 

 

Have you done your TPS reports?

 

From: AF mailto:af-boun...@af.afmug.com> > On
Behalf Of Steve Jones
Sent: Wednesday, February 12, 2025 3:51 PM
To: AnimalFarm Microwave Users Group mailto:af@af.afmug.com> >
Subject: [AFMUG] hired manager

 

so I've been putting my poker in a lot of fires lately.

 

in a couple instances in an upperish managerial role for small companies

 

I'm finding more idle time than expected in regard to the management aspect
of the roles. the other unrelated tasks fill the gaps, but when that's done

 

does management actually do stuff through the whole day?

 

I haven't had a single role position in decades.

 

is this why fractional employment is so popular now?

 

it seems everything is always waiting on something, a call, a meeting, a
task completion by somebody else.

-- 
AF mailing list
AF@af.afmug.com
http://af.afmug.com/mailman/listinfo/af_af.afmug.com

Re: [AFMUG] DHCP Fail over

[Sterling Jacobson via AF]

Dennis, isn't that a recipe for double IP assignments?

Wouldn't each DHCP server (relay DHCP endpoint server) need to have 
non-overlapping IPv4 pools?

ASFAIK there is no actual HA replication of DHCP tables on a server, so if one 
server is always responding to the layer2 domain request and it becomes 
unavailable the secondary or tertiary server would answer with a stale table 
and possibly assign a duplicate?


From: AF  on behalf of Dennis Burgess - LTI Support 
via AF 
Sent: Wednesday, February 12, 2025 1:52 PM
To: AnimalFarm Microwave Users Group 
Cc: Dennis Burgess - LTI Support 
Subject: Re: [AFMUG] DHCP Fail over


You can also run a mikrotik DHCP Server with relay going to each server, the MT 
server can run virtually and have high availability on itself, but the three 
DHCP (relays), will all be pulling from the same pool.



From: AF  On Behalf Of Josh Luthman
Sent: Tuesday, February 11, 2025 6:09 PM
To: AnimalFarm Microwave Users Group 
Subject: Re: [AFMUG] DHCP Fail over



Kea is what you want, I think...



https://www.isc.org/kea/



For HA: 
https://kea.readthedocs.io/en/latest/arm/hooks.html#supported-configurations



On Tue, Feb 11, 2025 at 6:23 PM Adam Moffett 
mailto:dmmoff...@gmail.com>> wrote:

We have two DHCP servers per market and they run VRRP.  VRRP gives you an 
active/standby setup.  Configurations have to be synchronized of course, but 
I'd say this is the simplest way.



To have any kind of active/active setup the DHCP servers would have to share 
the same lease database.  I believe ISC had a way to do that where they would 
send messages to update each other, but I haven't looked into this in awhile so 
I may be hallucinating that.



-Adam







From: AF mailto:af-boun...@af.afmug.com>> on behalf of 
Jesse DuPont 
mailto:jesse.dup...@celeritycorp.net>>
Sent: Tuesday, February 11, 2025 5:29 PM
To: AnimalFarm Microwave Users Group mailto:af@af.afmug.com>>
Subject: Re: [AFMUG] DHCP Fail over



What will you be using for your DHCP "concentrator" (for lack of a better 
term); that is, what will be the gateway device(s)? It seems you'd be better 
served by having a pair of routers running VRRP or some other cluster prototol, 
then having redundant DHCP servers that the concentrators/gateways relay to 
simultaneously (both of which check with RADIUS for auth and assignment for 
statics). The two DHCP servers can be configured active/active or active/backup 
and they'll both serve the same blocks (based on what RADIUS tells them to 
provide). ISC DHCP did this "okay", but KEA DHCP (ISC's replacement) does it 
really well. The two gateways using VRRP would appear like a single device and 
have a single IP. Depending on the routers, sometimes "state" (like current ARP 
resolutions) are sync'd between both routers, sometimes the failover router has 
to just re-ARP for everything; not the end of the world.

You can simplify all this by using an actual BNG for your DHCP side (and your 
PPPoE, for that matter). Something like NetElastic's or IP Infusion's BNG can 
do all this.

On 2/11/25 3:12 PM, Mark - Myakka Technologies via AF wrote:

We currently run 3 PPPoE servers using an OSPF concentrator and radius to 
manage the IP addresses.  With this setup, it doesn't matter which IP lands on 
which PPPoE server.  OSFP handles it.



We now need to do something similar with DHCP.  I've been messing around with 
/32's and Option 121, but just can not get a stable solution.  I'm now thinking 
about plan B.  Similar general setup we use on the PPPoE side.  Lets say we go 
with 3 DHCP servers connected to an OSPF concentrator.  I would have to set my 
DHCP network on all 3 servers to something like 
192.168.0.0/23 for about 512 address total.  Server one 
will do a GW of 192.168.0.1, Server two will do a GW of 192.168.0.2, server 3 
will do a GW of 192.168.0.3.  When a client connects they will randomly connect 
to one of the 3 servers and receive an IP address from radius.  My current 
thoughts are



1. Each server will have a /32 address not the /23.  IP address on server 1 
will be 192.168.0.1/32.

2. OSFP will only announce the /32 address of the server to the concentrator.

3. I will have to use the DHCP script option to insert and delete the clients 
ip address as a /32 in OSPF on the server to update the concentrator.



The one issue I see off the bat is when a client reboots.  If the client 
reboots and moves from server 1 to server 3, I now have two servers with the 
same IP address.  I think I can deal with that by using a short lease time.





Thoughts?  I'm still digging around looking for other (better) options of 
having DHCP fail-over.  The one option that will not work is reserving a block 
of IPs per server.  We have several customers that are using static IPs, so 
they need to be accessible from all 3 servers.





--



Thanks,

 Mark   

Re: [AFMUG] hired manager

[Sterling Jacobson via AF]

Ummm, I'm going to need you to come in on Sunday...

Steve, you are right about your observations from my limited experience.
That is why fractional is becoming more of an option, and it should be.

It's also why management needs more bodies to fill their time managing them lol


From: AF  on behalf of Ken Hohhof 
Sent: Wednesday, February 12, 2025 3:05 PM
To: 'AnimalFarm Microwave Users Group' 
Subject: Re: [AFMUG] hired manager


Have you done your TPS reports?



From: AF  On Behalf Of Steve Jones
Sent: Wednesday, February 12, 2025 3:51 PM
To: AnimalFarm Microwave Users Group 
Subject: [AFMUG] hired manager



so I've been putting my poker in a lot of fires lately.



in a couple instances in an upperish managerial role for small companies



I'm finding more idle time than expected in regard to the management aspect of 
the roles. the other unrelated tasks fill the gaps, but when that's done



does management actually do stuff through the whole day?



I haven't had a single role position in decades.



is this why fractional employment is so popular now?



it seems everything is always waiting on something, a call, a meeting, a task 
completion by somebody else.
-- 
AF mailing list
AF@af.afmug.com
http://af.afmug.com/mailman/listinfo/af_af.afmug.com

[AFMUG] hired manager

[Steve Jones]

so I've been putting my poker in a lot of fires lately.

in a couple instances in an upperish managerial role for small companies

I'm finding more idle time than expected in regard to the management aspect
of the roles. the other unrelated tasks fill the gaps, but when that's done

does management actually do stuff through the whole day?

I haven't had a single role position in decades.

is this why fractional employment is so popular now?

it seems everything is always waiting on something, a call, a meeting, a
task completion by somebody else.
-- 
AF mailing list
AF@af.afmug.com
http://af.afmug.com/mailman/listinfo/af_af.afmug.com

Re: [AFMUG] hired manager

[Ken Hohhof]

Have you done your TPS reports?

 

From: AF  On Behalf Of Steve Jones
Sent: Wednesday, February 12, 2025 3:51 PM
To: AnimalFarm Microwave Users Group 
Subject: [AFMUG] hired manager

 

so I've been putting my poker in a lot of fires lately.

 

in a couple instances in an upperish managerial role for small companies

 

I'm finding more idle time than expected in regard to the management aspect of 
the roles. the other unrelated tasks fill the gaps, but when that's done

 

does management actually do stuff through the whole day?

 

I haven't had a single role position in decades.

 

is this why fractional employment is so popular now?

 

it seems everything is always waiting on something, a call, a meeting, a task 
completion by somebody else.

-- 
AF mailing list
AF@af.afmug.com
http://af.afmug.com/mailman/listinfo/af_af.afmug.com

Re: [AFMUG] DHCP Fail over

[Dennis Burgess - LTI Support via AF]

No, they would not have pools, they would be running VRRP, but DHCP server 
(relayee) is behind it..  it has the block..

From: AF  On Behalf Of Sterling Jacobson via AF
Sent: Wednesday, February 12, 2025 3:42 PM
To: AnimalFarm Microwave Users Group 
Cc: Sterling Jacobson 
Subject: Re: [AFMUG] DHCP Fail over

Dennis, isn't that a recipe for double IP assignments?

Wouldn't each DHCP server (relay DHCP endpoint server) need to have 
non-overlapping IPv4 pools?

ASFAIK there is no actual HA replication of DHCP tables on a server, so if one 
server is always responding to the layer2 domain request and it becomes 
unavailable the secondary or tertiary server would answer with a stale table 
and possibly assign a duplicate?


From: AF mailto:af-boun...@af.afmug.com>> on behalf of 
Dennis Burgess - LTI Support via AF mailto:af@af.afmug.com>>
Sent: Wednesday, February 12, 2025 1:52 PM
To: AnimalFarm Microwave Users Group mailto:af@af.afmug.com>>
Cc: Dennis Burgess - LTI Support 
mailto:dmburg...@linktechs.net>>
Subject: Re: [AFMUG] DHCP Fail over


You can also run a mikrotik DHCP Server with relay going to each server, the MT 
server can run virtually and have high availability on itself, but the three 
DHCP (relays), will all be pulling from the same pool.



From: AF mailto:af-boun...@af.afmug.com>> On Behalf Of 
Josh Luthman
Sent: Tuesday, February 11, 2025 6:09 PM
To: AnimalFarm Microwave Users Group mailto:af@af.afmug.com>>
Subject: Re: [AFMUG] DHCP Fail over



Kea is what you want, I think...



https://www.isc.org/kea/



For HA: 
https://kea.readthedocs.io/en/latest/arm/hooks.html#supported-configurations



On Tue, Feb 11, 2025 at 6:23 PM Adam Moffett 
mailto:dmmoff...@gmail.com>> wrote:

We have two DHCP servers per market and they run VRRP.  VRRP gives you an 
active/standby setup.  Configurations have to be synchronized of course, but 
I'd say this is the simplest way.



To have any kind of active/active setup the DHCP servers would have to share 
the same lease database.  I believe ISC had a way to do that where they would 
send messages to update each other, but I haven't looked into this in awhile so 
I may be hallucinating that.



-Adam







From: AF mailto:af-boun...@af.afmug.com>> on behalf of 
Jesse DuPont 
mailto:jesse.dup...@celeritycorp.net>>
Sent: Tuesday, February 11, 2025 5:29 PM
To: AnimalFarm Microwave Users Group mailto:af@af.afmug.com>>
Subject: Re: [AFMUG] DHCP Fail over



What will you be using for your DHCP "concentrator" (for lack of a better 
term); that is, what will be the gateway device(s)? It seems you'd be better 
served by having a pair of routers running VRRP or some other cluster prototol, 
then having redundant DHCP servers that the concentrators/gateways relay to 
simultaneously (both of which check with RADIUS for auth and assignment for 
statics). The two DHCP servers can be configured active/active or active/backup 
and they'll both serve the same blocks (based on what RADIUS tells them to 
provide). ISC DHCP did this "okay", but KEA DHCP (ISC's replacement) does it 
really well. The two gateways using VRRP would appear like a single device and 
have a single IP. Depending on the routers, sometimes "state" (like current ARP 
resolutions) are sync'd between both routers, sometimes the failover router has 
to just re-ARP for everything; not the end of the world.

You can simplify all this by using an actual BNG for your DHCP side (and your 
PPPoE, for that matter). Something like NetElastic's or IP Infusion's BNG can 
do all this.

On 2/11/25 3:12 PM, Mark - Myakka Technologies via AF wrote:

We currently run 3 PPPoE servers using an OSPF concentrator and radius to 
manage the IP addresses.  With this setup, it doesn't matter which IP lands on 
which PPPoE server.  OSFP handles it.



We now need to do something similar with DHCP.  I've been messing around with 
/32's and Option 121, but just can not get a stable solution.  I'm now thinking 
about plan B.  Similar general setup we use on the PPPoE side.  Lets say we go 
with 3 DHCP servers connected to an OSPF concentrator.  I would have to set my 
DHCP network on all 3 servers to something like 
192.168.0.0/23 for about 512 address total.  Server one 
will do a GW of 192.168.0.1, Server two will do a GW of 192.168.0.2, server 3 
will do a GW of 192.168.0.3.  When a client connects they will randomly connect 
to one of the 3 servers and receive an IP address from radius.  My current 
thoughts are



1. Each server will have a /32 address not the /23.  IP address on server 1 
will be 192.168.0.1/32.

2. OSFP will only announce the /32 address of the server to the concentrator.

3. I will have to use the DHCP script option to insert and delete the clients 
ip address as a /32 in OSPF on the server to update the concentrator.



The one issue I see off the bat is when a client reboots.  If the client 
reboo