Onboarding a switch with preexisting config

[a_stan]

I have a Dell switch in production. I now need to add it to HiveManager. Must I erase the startup config, reboot the switch and configure it from scratch in HM interface? Or is there a less intrusive way for me to perform this?

[martin_ryan]

Hi BJ,

Assigning port types, VLANs, etc via templates is no problem and we do the same.

However, using the channel-group, sub-interface command, you can create a port channel - a new interface in it's own right. Where in hive manager are you going to configure the characteristics of this new interface? Depending on your needs, a given vlan may include a layer-3 interface which on the CLI, you bring into existence by simply declaring, "interface vlanN". Again, where in hive manager are you going to store the ip address and other characteristics of this interface?

Some answer with "the supplemental CLI" but what happens then? Every time you push a config update - say, changing the vlan a normal, switched port is in - are those supplemental commands executed? Are they looked for in the config and executed if they're missing? How does hive manager ensure that at least one port has been configured with "channel-group" before executing the supplemental commands that presume the existence of a port channel interface? Or, does it assume they pre-exist and execute the port channel commands blindly?

On the CLI, referring to "interface poN" will not be accepted unless you've created the port channel first. How this dependency is expressed in hive manager is far from clear. As a result, in our case, hive manager has been reduced to a glorified equipment monitor with all configuration updates handled by some other means.

[a_stan]

It is an N-3000 switch. Eventually I gave up. I was successful adding 2 N-2000 switches in a stack. No outage experienced during the tests. When pushing config from HM to the stack with pre-existing config, it seems to be merging the 2 configurations; I thought it would first wipe it clean and then apply new config. Where the 2 configs conflict - the one pushed from HM is applied on the switch. It all went very smooth.

Still not clear what went wrong with N3000; HM seems to have changed to vty password so I can no longer access the switch (as the console port is broken). It appears online in HM, but I can't change the config on it. ¯_(ツ)_/¯

[andrew_vernon]

I know this thread's a couple of months old, but I can consistently recreate your problem on my Dell N2000 series switches. I've found that doing "erase startup-config" on the switch and treating it as new allows HM to configure it, but starting from scratch isn't always an option.

Which switch family are you using? I know N2000 has issues. I've got a couple of new N1100's for testing, so I'm interested in seeing if it's common across Dell platforms.

I also have SR2208P's around, but since they're Aerohive switches, they were on HiveManager from unboxing.

[martin_ryan · Accepted solution]

Hi Ana-Maria,

Be warned - I'm new to Dell switches in HiveManager as well - bit Idid have a similar issue. Unfortunately, I tried a number of things to resolve it at the same time so, I'm not sure which solved it - but two things to check;

1) I've couldn't see where routed ports and other layer3 information is to be stored in HiveManager - so, just check that the config push is not "cutting off the branch you're sitting on" - i.e. that the interface you need for HiveManager to reach the switch is not being cut off by inappropriate config.

2) I had AAA running on the switch and one of the issues was that there were authentication failures on the ACS server for user "hivemanager" (or something similar. Ensuring local authentication works and turning off AAA authen and authorization may help.

Best of luck, Martin