VRRP- Need lots of clarification

[jenns]

Hello,

I'm new to implementing VRRP and A/V streams and need a lot of help here with a situation. I have 3x Cisco 9300s. Attached you'll see that I have a primary and secondary switch connected to another switch that contains a video server and a VLC client on the same VLAN.  My goal is for the client to pull streams with failover capability from the two switches on the left. On Sw 1 Pri and Sw2 Backup, I have the following configurations on VLAN100:

Sw1 PRI:

interface Vlan100
description Test
ip address x.x.100.254 255.255.255.0
vrrp 100 address-family ipv4
priority 110
address x.x.100.1 primary
exit-vrrp

SW2 Backup:

interface Vlan100
description Test
ip address x.x.100.253 255.255.255.0
vrrp 100 address-family ipv4
priority 105
address x.x.100.1 primary
exit-vrrp 

When I start the stream, the VLC can pull it just fine. When I initiate a failover by unplugging the trunk connection, I can see that SW2 Backup logs state that it is taking over as master. This is also evidenced in my sh vrrp command. So as far as I can tell from the logs, it is working as intended. My problem is that it is not reflected in the stream. Regardless of if I'm failing the primary to backup or vice versa, the VLC stream is seamless and not interrupted at all. My leadership has bought up the question that if VRRP is working, the stream should at least see some small form of interrupt.  Has anyone had experience with a similar setup? Is the stream still supposed to experience a drop regardless of VRRP implementation? Thank you.

[balaji.bandi]

as per the diagram where is the client connected ? who is using VLC to get stream,

the goal of VRRP and HSRP or any other load balance protocol to give non-interupt services ?

Try to understand the issue, is the Steam not breaking is the issue ? or you expecting to break ?

Most of the streaming system, they keep buffering some  - you can notice even youtube, even you lost connection of Internet still the video play for sometime until the buffer in place.

[jenns]

Hello, thank you for responding!

I have a PI connected to the third switch that is using VLC to live stream. I understand the concept of VRRP being a failover as to protect services from interrupting, but I am specifically trying to understand if I should be seeing any sort of at least minor pause while the network is being switched over from primary to backup. As for the youtube-- that is a good point and I see what you mean, but would that apply here to a live stream?

[paul driver · Accepted solution]

Hello
If this is MC streaming and the source/receiver are within the same vlan then the VRRP failover even isn't applicable as all the mc traffic is locally significant via igmp, and even if the MC is source/receivers are on separate vlans, again as long a the pim/routing path have parity which should be because upon vrrp failover this shouldn't have change you wont see any interrupt of service. 

[mlund]

It looks like both client and the server is in vlan100. That means that communication between these 2 does not use the core switches for communication. Instead the communication is local layer2 in the switch3. That's why there is no communication break when you disconnect the core switch.

Vrrp is used for gateway redundancy, that means, when traffic is intended not for local subnet, but for any other subnet, that traffic is sent to the default-gateway. 

[balaji.bandi]

Sure in same VLAN it go in L2 broadcast. if you try different subnet, run some Wireshark and see you can get any, make sure if not same VLAN check TTL to be increased.

[paul driver]

Hello

---

@jenns wrote:

Hello,

Regardless of if I'm failing the primary to backup or vice versa, the VLC stream is seamless and not interrupted at all

---

This is what is expected, vrrp is used for redundancy for your lan clients that associate with the vrrp group for their respective vlan They wont be aware of any failover they as they point to the virtual ip/mac address of the vrrp group for their first next hop ip address (default-gateway) so as/when vrrp failover occurs the end hosts wont be affected.

[jenns]

Hi everyone!

Thanks for answering, now I have an update. I have put hosts in different VLANS so that they do not rely on their internal network for streaming and must rely on VRRP from my L3 switches. I ran the tests again, and still no visible interrupt on Wireshark or stream.  As some as you have helpfully noted the hosts should automatically use the virtual IP as next hop so they will not sense any drop in stream.

This leads to some more questions from me: Is the virtual ID meant to be pingable from the hosts? I have read some literature stating that it may be completely operational but not pingable depending on your setup and configuration. Does anyone have any knowledge on this? If I can just show it in my ARP table, then that should be okay right? 

Second question: When I fail from primary to backup, I see an automatic failover in the backup in VRRP states. When I plug the link back into the master, the states take much longer to revert back. My theory is that it's because VRRP is designed to account for failover to quickly prevent interrupt of service, but not necessarily accounting for quickly reverting when the master comes up. Does this sound correct or am I totally off the mark?

[balaji.bandi]

you should able to ping VIP IP sure.

depends on the config and it may be default value if you do not mentioned timers

that can be tweaked based on the requirement: check some example config ( ignore model)

https://www.cisco.com/c/en/us/td/docs/routers/asr9000/software/asr9k_r4-0/addr_serv/configuration/guide/ic40asr9kbook_chapter10.html