Understanding the GLBP Debug Messages and Timer

vijay.swaminathan · ‎05-04-2012

Hi Experts,

would need some clarity on GLBP : I have the topology as R1 -----R 2-------R3:

R3 is the AVG, R2 is standby AVG and R1 is in Listen state:

The output , and the configuration of R1, R2 and R3 are as follows:

R1:

===

interface FastEthernet0/0

ip address 10.10.123.1 255.255.255.0

glbp 1 ip 10.10.123.100

glbp 1 priority 110

glbp 1 preempt

glbp 1 weighting 10

glbp 1 load-balancing weighted

R1#sh glbp brief

Interface Grp Fwd Pri State Address Active router Standby router

Fa0/0 1 - 110 Listen 10.10.123.100 10.10.123.3 10.10.123.2

Fa0/0 1 1 - Active 0007.b400.0101 local -

Fa0/0 1 2 - Listen 0007.b400.0102 10.10.123.2 -

Fa0/0 1 3 - Listen 0007.b400.0103 10.10.123.3 -

R2:

====

interface FastEthernet0/0

ip address 10.10.123.2 255.255.255.0

glbp 1 ip 10.10.123.100

glbp 1 priority 120

glbp 1 preempt

glbp 1 weighting 8

glbp 1 load-balancing weighted

R2#sh glbp brief

Interface Grp Fwd Pri State Address Active router Standby router

Fa0/0 1 - 120 Standby 10.10.123.100 10.10.123.3 local

Fa0/0 1 1 - Listen 0007.b400.0101 10.10.123.1 -

Fa0/0 1 2 - Active 0007.b400.0102 local -

Fa0/0 1 3 - Listen 0007.b400.0103 10.10.123.3 -

R3:

===

R3#sh run int fa0/0

interface FastEthernet0/0

ip address 10.10.123.3 255.255.255.0

duplex auto

speed auto

glbp 1 ip 10.10.123.100

glbp 1 priority 130

glbp 1 preempt

glbp 1 weighting 1

glbp 1 load-balancing weighted

R3#sh glbp brief

Interface Grp Fwd Pri State Address Active router Standby router

Fa0/0 1 - 130 Active 10.10.123.100 local 10.10.123.2

Fa0/0 1 1 - Listen 0007.b400.0101 10.10.123.1 -

Fa0/0 1 2 - Listen 0007.b400.0102 10.10.123.2 -

Fa0/0 1 3 - Active 0007.b400.0103 local -

R3#

My Understanding is:

1. AVG (R3) since AVG receives the Broadcast ARP request from the client and answers with the VMAC based on the Load Balancing algorithm configured (in this case weighted).

2. as per the diagram, R3 is responsible for VMAC 0007.b400.0103 , R2 for 0007.b400.0102 and R1 for 0007.b400.0101

Now Testing:

When the AVG fails:

************************

When the AVG fails, then R2 takes over as Active.

I'm trying to understand the process behind this from the debug messages:

ON R2:

R2#

*Mar 1 01:37:54.571: GLBP: Fa0/0 1.3 Ignoring Hello (135/10.10.123.1 < 167/10.10.123.3)

*Mar 1 01:37:54.591: GLBP: Fa0/0 1 Standby: g/Active timer expired (10.10.123.3)

*Mar 1 01:37:54.591: GLBP: Fa0/0 1 Active router IP is local, was 10.10.123.3

*Mar 1 01:37:54.591: GLBP: Fa0/0 1 Standby router is unknown, was local

*Mar 1 01:37:54.591: GLBP: Fa0/0 1 Standby -> Active

*Mar 1 01:37:54.591: %GLBP-6-STATECHANGE: FastEthernet0/0 Grp 1 state Standby -> Active

*Mar 1 01:37:54.595: GLBP: Fa0/0 1.3 Listen: g/Active timer expired

*Mar 1 01:37:54.595: GLBP: Fa0/0 1.3 Listen -> Active

R2#

*Mar 1 01:37:54.595: %GLBP-6-FWDSTATECHANGE: FastEthernet0/0 Grp 1 Fwd 3 state Listen -> Active

R2#

*Mar 1 01:38:04.627: GLBP: Fa0/0 1 Standby router is 10.10.123.1

R2#

My understanding and Questions are: The numbers represent the line of debug msgs

1. R2 received hello from R1 and compared the priority.

2. Active Timer expired. that is it did not hear from Active.

3. Now it has taken the role of Active. it was R3 earlier

4. Standby is still unknown (calculation in progress)

5 & 6. Standby to Active Transitions.

7, 8 and 9. What is this message telling ???? Is this about the AVF ??

Debug Messages on R1:

==================

R1#

*Mar 1 01:37:55.703: GLBP: Fa0/0 1 Listen: g/Active timer expired (10.10.123.3)

*Mar 1 01:37:55.703: GLBP: Fa0/0 1 Active router IP is unknown, was 10.10.123.3

*Mar 1 01:37:55.703: GLBP: Fa0/0 1 Listen -> Speak

*Mar 1 01:37:55.707: GLBP: Fa0/0 1.3 Listen: g/Active timer expired

*Mar 1 01:37:55.707: GLBP: Fa0/0 1.3 Listen -> Active

*Mar 1 01:37:55.707: %GLBP-6-FWDSTATECHANGE: FastEthernet0/0 Grp 1 Fwd 3 state Listen -> Active

*Mar 1 01:37:55.759: GLBP: Fa0/0 1 Active router IP is 10.10.123.2

*Mar 1 01:37:55.763: GLBP: Fa0/0 1 Standby router is unknown, was 10.10.123.2

*Mar 1 01:37:55.763: GLBP: Fa0/0 1.3 Active: i/Hello rcvd from higher pri Active router (135/10.10.123.2)

*Mar 1 01:37:55.763: GLBP: Fa0/0 1.3 Active -> Listen

R1#

*Mar 1 01:37:55.767: %GLBP-6-FWDSTATECHANGE: FastEthernet0/0 Grp 1 Fwd 3 state Active -> Listen

R1#

*Mar 1 01:38:03.183: %DUAL-5-NBRCHANGE: IP-EIGRP(0) 100: Neighbor 10.10.123.3 (FastEthernet0/0) is down: holding time expired

R1#

*Mar 1 01:38:05.703: GLBP: Fa0/0 1 Speak: f/Standby timer expired (unknown)

*Mar 1 01:38:05.703: GLBP: Fa0/0 1 Standby router is local

*Mar 1 01:38:05.703: GLBP: Fa0/0 1 Speak -> Standby

R1#

My understanding as per debug lines:

1 & 2. Active Timer Expired. (lost connection with AVG)

3. Now from AVG listen state it starts moving to speak state

4 - 11 . My understanding is: from Listen it goes into Active AVG then it hears about a higher priority from R2 and then goes back from Active back again to Listen.

12. When R2 (earlier standby) moved to active, then standby hold timer would have kicked off and the line 12 talks about the expiry of that time.

13 - 14 : This finally tells who the standby is:

Another question is:

for the VMAC 0007.b400.0103 for which R3 was active, somebody should take over as active for this VMAC. it should be either R1 or R2.. on what basis is this decision made ??

and where are those pertinent messages???

please advice:

Thanks

-Vijay

vijay.swaminathan · ‎05-04-2012

Any thoughts on this please??

-vijay

Giuseppe Larosa · ‎05-05-2012

Hello Vijay,

my understanding is different.

R2 debug lines:

R2 becomes new AVG

*Mar 1 01:37:54.591: %GLBP-6-STATECHANGE: FastEthernet0/0 Grp 1 state Standby -> Active

R2 takes over as AVF for forwarder 3 (that was performed by now failed R3) but R1 is the first to attempt to take this role.

*Mar 1 01:37:54.571: GLBP: Fa0/0 1.3 Ignoring Hello (135/10.10.123.1 < 167/10.10.123.3)

here R3 has not expired on R2 but R1 has already sent out its own hello that is ignored.

R2 lines

*Mar 1 01:37:54.595: GLBP: Fa0/0 1.3 Listen: g/Active timer expired

*Mar 1 01:37:54.595: GLBP: Fa0/0 1.3 Listen -> Active

R2#

R3 has expired and R2 promotes itself as AVF for VMAC #3

R1 debug lines:

R1 promotes itself to AVF for VMAC#3 for some time and then it gives up when receives better hello from R2

R1 never attempts to become the new AVG it moves to Speak state and later to Standby state.

*Mar 1 01:37:55.707: %GLBP-6-FWDSTATECHANGE: FastEthernet0/0 Grp 1 Fwd 3 state Listen -> Active

*Mar 1 01:37:55.763: GLBP: Fa0/0 1.3 Active: i/Hello rcvd from higher pri Active router (135/10.10.123.2)

*Mar 1 01:37:55.767: %GLBP-6-FWDSTATECHANGE: FastEthernet0/0 Grp 1 Fwd 3 state Active -> Listen

All debug messages with 1.3 refer to AVF activity for GLBP group 1 VMAC#3 as you have noted in your other thread.

So there are messages about a competition between R2 and R1 for taking over the AVF role that was performed by R3.

R1 was the first to detect the failure and claimed AVF role for VMAC#3 until after expiration of timer on R2 R2 starts to send out hellos claiming to be new AVF3

About AVG: there is no competition as R2 was the standby AVG (state Standby) is promoted to AVG state after R3 failure is detected

*Mar 1 01:37:54.591: GLBP: Fa0/0 1 Standby: g/Active timer expired (10.10.123.3)

*Mar 1 01:37:54.591: GLBP: Fa0/0 1 Standby -> Active

What I have found really interesting is that GLBP has a two dimensions "priority": priority and weighting.

In debug messages we see a single pri value associated to devices but it is not equal to configured priority:

Example:

*Mar 1 01:37:54.571: GLBP: Fa0/0 1.3 Ignoring Hello (135/10.10.123.1 < 167/10.10.123.3)

but R1 configured priority is 110 with weighting 10

and R3 configured priority is 130 with weighting 1

it is not clear how values 135 and 167 are calculated.

to be noted later the hello from R2 has pri 135:

*Mar 1 01:37:55.763: GLBP: Fa0/0 1.3 Active: i/Hello rcvd from higher pri Active router (135/10.10.123.2)

All we can say is that for AVG there is no competition among surviving nodes in GLBP group as AVG Standby is promoted and a new AVG Standby is elected (similar to OSPF reactions to DR failure BDR is promoted)

For AVF takeover a competition may happen between surviving nodes depending on expiration of timer related to failed AVF.

To work correctly in a LAN switched environment hello messages related to a specific AVF should have source MAC address = AVF MAC address in order to update CAM tables in involved switches

Hope to help

Giuseppe

vijay.swaminathan · ‎05-08-2012

Thanks Giuseppe,

That definately brought some insight to me.. yes.. you are right I still did not get to know how come R2 took over the active role for the VMAC 3 (which was earlier performed by R3) ..

to be precise I could not get what the highligted text means as you said

Fa0/0 1.3 Ignoring Hello (135/10.10.123.1 < 167/10.10.123.3)

What is this actually looking at when making decision of which switch takes the active role for the VMAC 3. As I can see that even though weight is less on R2 when compared to R1 , R2 takes up the active role and this is what Is confusing me :-(

-Vijay