Add new Member in IRF Stack

[Neonium]

Hi,

I currently have an IRF ring of 4 5820 switches. I would add 2 more 5820 switches now. I have the IRF port configuration made on the two new switches. So renamed 5 and 6 and the IRF port defined groups. Now I have this morning tried the two in the present IRF stack to connect, but nothing happened. I then restarted the present stack. Result is that nothing happens.
Then I read this must be the same firmware version on it. So I have on the two new switches have the same firmware on it as made on the existing stack, but it happened nothing. We use 4x10Gbit lines for an IRF port group. I had, but only 2 are connected by 4 wires. Ending with the ports then flashed orange.

Is there somewhere a guide how to present grabs two more switches in a or can any one tell me what am I doing wrong?

[Apachez-]

It should be the same procedure as when you add a new device to your IRF.

First check that you have the latest firmware available (since there are some developments and bugfixes even regarding IRF), then make sure all units have the same firmware.

Then when you connect it this method should work:

http://bizsupport1.austin.hp.com/bc/docs/support/SupportManual/c02648772/c02648772.pdf

http://allhpnetworking.blogspot.se/2010/10/hp-networking-series-example.html

http://blog.yogi-way.lv/2011/10/how-to-replace-h3c-stack-member.html

http://h20566.www2.hp.com/portal/site/hpsc/template.PAGE/public/kb/docDisplay/?sp4ts.oid=5195279&spf_p.tpst=kbDocDisplay&spf_p.prp_kbDocDisplay=wsrp-navigationalState%3DdocId%253Demr_na-c03239130-3%257CdocLocale%253D%257CcalledBy%253D&javax.portlet.begCacheTok=com.vignette.cachetoken&javax.portlet.endCacheTok=com.vignette.cachetoken

[Thej0sepH]

Hello,

First of all, be sure that your master has the highest IRF Priority so the new switches can't make the old ones lose their config.

Second, check if you have configured this:

1. Check that IRF Port 1 of the old switch connects with IRF Port2 in the new switches and vice versa.

2. Remember the command irf-port-configuration active, this is the one that makes all work.

Hope this helps

Greetings

[Vince-Whirlwind]

Definitely check your existing master has the highest priority first.

Have you even established you have Layer1 & Layer2 connectivity between the existing switches and the new ones?

[Neonium]

in the current stack, a switch has priority 32 and all the others have the priority 1 The two new switches have both the priority 1

On all switches, the firmware is 1808P23. I think I've found the error or is it connected no matter at what is Core1 or the sequence 1 must -> 2 -> 3 -> 4 -> 5 -> 6 -> 1 are complied with?

[Vince-Whirlwind]

Issue a "Display IRF topology" command.

Also, describe the physical topology (preferably with a diagram) including the Switchport IDs in use.

[Neonium]

then we have probably a mistake in the wiring. Because if Switch 1, the neighbors must have 2 and 6. Do I have to turn the cable. That you can not make it during running operations, without which there is likely failures.

[Vince-Whirlwind]

Again, we don't know what your topology is, but if you have it in a ring, as you should, then you can disconnect the physical interfaces associated with the IRF ports on either side of any one switch-switch link.

You need to shutdown the interfaces before configuring them into an IRF port.

Others can correct me if I am wrong, but the only reason you would want your switch numbering to follow any kind of sequence is for your own peace of mind - the switches aren't going to care.

If you've got something wrong, it's more likely to be:

 - single IRF port has multiple physical members patched to different neighbours

 - IRF port 1 patched to IRF port 1 on its neighbour (or 2 to 2)

[Neonium]

Yes. At the moment, the topology is correct, but if I were 5 and 6 to include the switches, would be the order
1 -> 4 -> 3 -> 5 -> 6 -> 2 ->

I just have to turn the line into 2 data centers now. At the moment the lines of Switch 1 run directly to another data center to Switch 4 To do it right would switch 1 run over the location where the switch 5 and 6. That's not to realize my opinion without failure. have ne time drawing made

[Peter_Debruyne]

Hi,

Always make sure the all members of the IRF stack have a different prio from 1.

Example:

switch1 32

switch2 30

switch3 28

switch4 26

etc.

Initially switch1 will be master, but if it has been rebooted (e.g. due to power failure), another switch would become master and remains master even when switch1 returns.

If you left the default for the other switches (prio 1), this would mean that your current master would be running with prio1, and could easily be overruled by a new device.

Also remember to use the irf-port-configuration active command every time you make a change to the IRF port configuration (so both on the existing irf system as on the newly joining devices).

This is because any change to the irf port configuration is always in a "pending"  state until you commit with the irf-port-configuration active command.

Finally, as mentioned above, check the physical links and respect the irf port 1 to irf port 2 rule,

hth,Peter