This is a bit more complicated than a simple test.

After extensive testing and timing everything... and changing the
hello-interval multiple times and retesting, I've come to a conclusion:
election takes place an average between three and four hello-internal times
after 2-way.  Again, I've tested this with many different
hello-interval values over the past few hours.

Not only that, but the theory about "highest routing ID wins" is no always
true.  I've got it to become the BDR.  I'd expect either DROTHER or DR.
 I've been able to recreate this numerous times:

(the following "ip addresses" are router IDs, not ip addresses)

**The following is CONSISTENT and REPEATABLE (it's easier to test with a
hello-internal of 5, so enable R3's interface at ~12 seconds after 2-way is
established**
If R3 (10.1.3.1) comes on too late (after election), it's a DROTHER.  That
makes sense since it's late to the party.
If R3 comes on with R1 (10.1.1.1) and R2 (10.1.2.1), it's the DR.
If R3 comes on 2-3 seconds before election is going to begin (time it
right), it's the BDR-- with  R2 (10.1.2.1) beating it, but it itself beating
R1 (10.1.1.1)

I'm satisfied with my results.


On Wed, Nov 24, 2010 at 1:20 PM, Jason Maynard <[email protected]>wrote:

> Antonio.
>
> Am I missing something (I want to make sure that I understand)
>
> from the output below  (single router running OSPF)
>
> >
> > *Mar 1 00:01:07.091: OSPF: Interface FastEthernet0/0 going Up  *(OSPF
> interface comes up)*
>
> >
> > *Mar 1 00:01:07.595: OSPF: Build router LSA for area 0, router ID
> > 1.1.1.1, seq 0x80000001
> >
> > *Mar 1 00:01:47.091: OSPF: end of Wait on interface FastEthernet0/0 *(Wait
> time ended)*
> >
> > *Mar 1 00:01:47.091: OSPF: DR/BDR election on FastEthernet0/0 *(Election
> begins)*
>
> Jason Maynard
>
>
> ------------------------------
> *From:* antonio tuozzo <[email protected]>
> *To:* Jason Maynard <[email protected]>
> *Cc:* David Betz <[email protected]>; [email protected]
> *Sent:* Wed, November 24, 2010 2:15:12 PM
>
> *Subject:* Re: [OSL | CCIE_RS] OSPF Election Time
>
> Jason,
> wait-timer = 40 sec = 4 x hello-timer (10 sec) = dead-interval on
> broadcast (ospf) network type.
>
> regard
> antonio
>
> >
> > *Mar 1 00:01:07.091: OSPF: Interface FastEthernet0/0 going Up
> >
> > *Mar 1 00:01:07.595: OSPF: Build router LSA for area 0, router ID
> > 1.1.1.1, seq 0x80000001
> >
> > *Mar 1 00:01:47.091: OSPF: end of Wait on interface FastEthernet0/0
> >
> > *Mar 1 00:01:47.091: OSPF: DR/BDR election on FastEthernet0/0
> >
> > Jason Maynard
> >
> >
> > ------------------------------------------------------------------------
> > *From:* David Betz <[email protected]>
> > *To:* [email protected]
> > *Sent:* Wed, November 24, 2010 9:53:35 AM
> > *Subject:* Re: [OSL | CCIE_RS] OSPF Election Time
> >
> > The results are not in question, the timing is.
> >
> > My current theory is that election is always from the list of routers
> > that a rouer has a two-way relationship with, thus the question has
> > nothing to do with that and needs to be pushed back to: "How long does a
> > router wait to collect two-way state neighbors before stop waiting and
> > starting and election?"
> >
> > On Wed, Nov 24, 2010 at 5:14 AM, Jason Maynard <[email protected]
> > <mailto:[email protected]>> wrote:
> >
> >    I have not read all the comments but I thought I would try this
> >
> >    I configured OSPF on a single router and ran “debug ip ospf adja”
> >
> >    *Mar 1 00:01:07.091: OSPF: Interface FastEthernet0/0 going Up
> >
> >    *Mar 1 00:01:07.595: OSPF: Build router LSA for area 0, router ID
> >    1.1.1.1, seq 0x80000001
> >
> >    *Mar 1 00:01:47.091: OSPF: end of Wait on interface FastEthernet0/0
> >
> >    *Mar 1 00:01:47.091: OSPF: DR/BDR election on FastEthernet0/0
> >
> >    *Mar 1 00:01:47.095: OSPF: Elect BDR 1.1.1.1
> >
> >    *Mar 1 00:01:47.095: OSPF: Elect DR 1.1.1.1
> >
> >    *Mar 1 00:01:47.099: OSPF: Elect BDR 0.0.0.0
> >
> >    *Mar 1 00:01:47.099: OSPF: Elect DR 1.1.1.1
> >
> >    *Mar 1 00:01:47.103: DR: 1.1.1.1 (Id)BDR: none
> >
> >    *Mar 1 00:01:47.603: OSPF: No full nbrs to build Net Lsa for
> >    interface FastEthernet0/0
> >
> >    Does this help?
> >
> >    *From:*[email protected]
> >    <mailto:[email protected]>
> >    [mailto:[email protected]
> >    <mailto:[email protected]>] *On Behalf Of *David
> Betz
> >    *Sent:* November-24-10 3:37 AM
> >
> >
> >    *To:* [email protected] <mailto:[email protected]
> >
> >    *Subject:* Re: [OSL | CCIE_RS] OSPF Election Time
> >
> >    That's my current theory. However, this really shouldn't be a
> >    theory, but scientific fact. I've looked at the RFC and still
> >    haven't found anything. Granted, it was a skim. I'd need to scan it
> >    to be sure of it's absence or presence.
> >
> >    On Wed, Nov 24, 2010 at 2:10 AM, antonio <[email protected]
> >    <mailto:[email protected]>> wrote:
> >
> >    Maybe you have to look at wait timer (by default equals to
> >    dead-interval)
> >
> >    regards
> >    antonio
> >
> >    Il 24/11/2010 03:43, David Betz ha scritto:
> >
> >    Upon closer examination of the process, it seems that DR/BDR election
> is
> >    based on the candidates that are adjacencies (neighbor routers with
> >    priority > 0). Thus, the question is pushed backed a step: when does a
> >    router say "I've 2-way with n-number of routers on this broadcast
> >    segment, let's begin election."? Even then, multiple routers are doing
> >    this, but given that that is per-segment, the election results should
> be
> >    the same all over. So, that part shouldn't be an issue.
> >
> >    On Tue, Nov 23, 2010 at 7:58 PM, David Betz <[email protected]
> >    <mailto:[email protected]>
> >
> >    <mailto:[email protected] <mailto:[email protected]>>> wrote:
> >
> >    Yes?
> >
> >    I've done this experiment myriad times and looked over the data for
> >    a long time. What is your personal analysis of this data? Any
> >    conclusion? I've not been any to come up with with anything based
> >    on this data. If we are going to do our work off empirical data, we
> >    need to have a series of tests with a specified baseline. However,
> >    I'm more looking for an axiom from the [seemingly non-existent]
> >    documentation. Normally I'd look at the RFC, but do not any any
> >    reason to think that Cisco is following it at this point.
> >
> >
> >    On Tue, Nov 23, 2010 at 7:49 PM, Jason Maynard
> >
> >    <[email protected] <mailto:[email protected]>
> >    <mailto:[email protected] <mailto:[email protected]>>>
> >    wrote:
> >
> >    Have a look at the times below. Between two routers in a
> >    broadcast network
> >
> >    *Mar 1 00:09:00.215: OSPF: end of Wait on interface FastEthernet0/0
> >
> >    *Mar 1 00:09:00.215: OSPF: DR/BDR election on FastEthernet0/0
> >
> >    *Mar 1 00:09:00.219: OSPF: Elect BDR 222.222.222.222
> >
> >    *Mar 1 00:09:00.219: OSPF: Elect DR 222.222.222.222
> >
> >    *Mar 1 00:09:00.223: OSPF: Elect BDR 0.0.0.0
> >
> >    *Mar 1 00:09:00.223: OSPF: Elect DR 222.222.222.222
> >
> >    *Mar 1 00:09:00.227: DR: 222.222.222.222 (Id) BDR: none
> >
> >    *Mar 1 00:09:00.323: OSPF: 2 Way Communication to 1.1.1.1 on
> >    FastEthernet0/0, state 2WAY
> >
> >    *Mar 1 00:09:00.323: OSPF: Neighbor change Event on interface
> >    FastEthernet0/0
> >
> >    *Mar 1 00:09:00.327: OSPF: DR/BDR election on FastEthernet0/0
> >
> >    *Mar 1 00:09:00.327: OSPF: Elect BDR 1.1.1.1
> >
> >    *Mar 1 00:09:00.331: OSPF: Elect DR 222.222.222.222
> >
> >    *Mar 1 00:09:00.331: DR: 222.222.222.222 (Id) BDR:
> >    1.1.1.1 (Id)
> >
> >    *Mar 1 00:09:00.331: OSPF: Send DBD to 1.1.1.1 on
> >    FastEthernet0/0 seq 0xA62 opt 0x52 flag 0x7 len 32
> >
> >    *Mar 1 00:09:00.331: OSPF: Neighbor change Event on interface
> >    FastEthernet0/0
> >
> >    *Mar 1 00:09:00.331: OSPF: DR/BDR election on FastEthernet0/0
> >
> >    *Mar 1 00:09:00.331: OSPF: Elect BDR 1.1.1.1
> >
> >    *Mar 1 00:09:00.331: OSPF: Elect DR 222.222.222.222
> >
> >    *Mar 1 00:09:00.331: DR: 222.222.222.222 (Id) BDR:
> >    1.1.1.1 (Id)
> >
> >    *Mar 1 00:09:00.331: OSPF: Neighbor change Event on interface
> >    FastEthernet0/0
> >
> >    *Mar 1 00:09:00.331: OSPF: DR/BDR election on FastEthernet0/0
> >
> >    *Mar 1 00:09:00.331: OSPF: Elect BDR 1.1.1.1
> >
> >    *Mar 1 00:09:00.331: OSPF: Elect DR 222.222.222.222
> >
> >    *Mar 1 00:09:00.331: DR: 222.222.222.222 (Id) BDR:
> >    1.1.1.1 (Id)
> >
> >    *Mar 1 00:09:00.331: OSPF: Rcv DBD from 1.1.1.1 on
> >    FastEthernet0/0 seq 0x1376 opt 0x52 flag 0x7 len 32 mtu 1500
> >    state EXSTART
> >
> >    *Mar 1 00:09:00.331: OSPF: First DBD and we are not SLAVE
> >
> >    *Mar 1 00:09:00.379: OSPF: Rcv DBD from 1.1.1.1 on
> >    FastEthernet0/0 seq 0xA62 opt 0x52 flag 0x2 len 52 mtu 1500
> >    state EXSTART
> >
> >    *Mar 1 00:09:00.383: OSPF: NBR Negotiation Done. We are the MASTER
> >
> >    *Mar 1 00:09:00.387: OSPF: Send DBD to 1.1.1.1 on
> >    FastEthernet0/0 seq 0xA63 opt 0x52 flag 0x3 len 52
> >
> >    *Mar 1 00:09:00.475: OSPF: Rcv DBD from 1.1.1.1 on
> >    FastEthernet0/0 seq 0xA63 opt 0x52 flag 0x0 len 32 mtu 1500
> >    state EXCHANGE
> >
> >    *Mar 1 00:09:00.479: OSPF: Send DBD to 1.1.1.1 on
> >    FastEthernet0/0 seq 0xA64 opt 0x52 flag 0x1 len 32
> >
> >    *Mar 1 00:09:00.483: OSPF: Send LS REQ to 1.1.1.1 length 12 LSA
> >    count 1
> >
> >    *Mar 1 00:09:00.567: OSPF: Rcv LS REQ from 1.1.1.1 on
> >    FastEthernet0/0 length 36 LSA count 1
> >
> >    *Mar 1 00:09:00.571: OSPF: Send UPD to 10.0.0.2 on
> >    FastEthernet0/0 length 40 LSA count 1
> >
> >    *Mar 1 00:09:00.619: OSPF: Rcv DBD from 1.1.1.1 on
> >    FastEthernet0/0 seq 0xA64 opt 0x52 flag 0x0 len 32 mtu 1500
> >    state EXCHANGE
> >
> >    *Mar 1 00:09:00.623: OSPF: Exchange Done with 1.1.1.1 on
> >    FastEthernet0/0
> >
> >    *Mar 1 00:09:00.623: OSPF: Rcv LS UPD from 1.1.1.1 on
> >    FastEthernet0/0 length 64 LSA count 1
> >
> >    *Mar 1 00:09:00.627: OSPF: Synchronized with 1.1.1.1 on
> >    FastEthernet0/0, state FULL
> >
> >    *Mar 1 00:09:00.631: %OSPF-5-ADJCHG: Process 1, Nbr 1.1.1.1 on
> >    FastEthernet0/0 from LOADING to FULL, Loading Done
> >
> >    *Mar 1 00:09:00.727: OSPF: Build network LSA for
> >    FastEthernet0/0, router ID 222.222.222.222
> >
> >    *Mar 1 00:09:00.731: OSPF: Build network LSA for
> >    FastEthernet0/0, router ID 222.222.222.222
> >
> >    *Mar 1 00:09:00.735: OSPF: Build router LSA for area 0, router
> >    ID 222.222.222.222, seq 0x80000004
> >
> >    *Mar 1 00:09:03.439: OSPF: Rcv LS UPD from 1.1.1.1 on
> >    FastEthernet0/0 length 64 LSA count 1
> >
> >    *From:*Jason Maynard [mailto:[email protected]
> >    <mailto:[email protected]>
> >    <mailto:[email protected] <mailto:[email protected]>>]
> >    *Sent:* November-23-10 8:25 PM
> >    *To:* 'David Betz'; '[email protected]
> >    <mailto:[email protected]>
> >
> >    <mailto:[email protected]
> >    <mailto:[email protected]>>'
> >
> >
> >    *Subject:* RE: [OSL | CCIE_RS] OSPF Election Time
> >
> >    Well that depends
> >
> >    How long does the device take to boot up compared to other devices
> >
> >    When is the device ready to participate in the OSPF election
> >    process (running more services on a device may take it longer to
> >    start participating compared to a router with less services)
> >
> >    DR and BDR election is done via the Hello protocol –
> >
> >    OSPF Network Type | Hello | DR/BDR |
> >    _________________________________________
> >    Broadcast | 10sec | Elects DR/BDR|
> >    _________________________________________
> >    NonBroadcast | 30sec | Elects DR/BDR|
> >    _________________________________________
> >    P2MP | 30sec | No DR/BDR |
> >    _________________________________________
> >    P2MP NonBroadcast | 30sec | No DR/BDR |
> >    _________________________________________
> >    P2P | 10sec | No DR/BDR |
> >
> >    If you want to control which devices become the DR/BDR you can
> >    use “*ip ospf priority” and set all devices that you do not want
> >    to be DR/BDR to “0” ***
> >
> >    Not sure exactly how may hellos before the DR/BDR election takes
> >    place, perhaps I will lab it up to see -
> >
> >    HTH
> >
> >    *From:*[email protected]
> >    <mailto:[email protected]>
> >    <mailto:[email protected]
> >    <mailto:[email protected]>>
> >    [mailto:[email protected]
> >    <mailto:[email protected]>
> >    <mailto:[email protected]
> >    <mailto:[email protected]>>] *On Behalf Of
> >    *David Betz
> >    *Sent:* November-23-10 7:32 PM
> >    *To:* [email protected] <mailto:[email protected]
> >
> >
> >    <mailto:[email protected]
> >    <mailto:[email protected]>>
> >
> >
> >    *Subject:* [OSL | CCIE_RS] OSPF Election Time
> >
> >    I've not been able to figure this out nor can I find any
> >    documentation for this: how long does OSPF wait for others to
> >    cast their vote before doing the DR/BDR election?
> >
> >    If R1, R2, and R3 come online with 10.1.1.X/25 (X=RX) ip
> >    addresses, and R1 and R2 see each other's messages, they will
> >    start an election. If R3 is there, it wins... if it's too late,
> >    R2 wins.
> >
> >    When is "too late"?
> >
> >
> >
> >
> >    _______________________________________________
> >    For more information regarding industry leading CCIE Lab training,
> >    please visit www.ipexpert.com <http://www.ipexpert.com>
> >
> >    _______________________________________________
> >    For more information regarding industry leading CCIE Lab training,
> >    please visit www.ipexpert.com <http://www.ipexpert.com>
> >
> >
> >
> >
> > _______________________________________________
> > For more information regarding industry leading CCIE Lab training, please
> visit www.ipexpert.com
>
_______________________________________________
For more information regarding industry leading CCIE Lab training, please visit 
www.ipexpert.com

Reply via email to