Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- [00:00] <Llama052_> give me a vlan map
- [00:00] <Llama052_> of all the vlans you have
- [00:00] <Llama052_> that need to ag into the untangle server
- [00:00] <Llama052_> also the unmanaged gigabit switch on the edge is strange
- [00:00] <Llama052_> nvm
- [00:01] <Llama052_> also you're showing 16x connections to the untangle server?
- [00:01] <Llama052_> on the specs
- [00:01] <Llama052_> Looks like you'll need to do vlan tagging on the untangle host
- [00:02] <Llama052_> and ignore vlan settings on the vmware box
- [00:02] <yancho> http://imgur.com/a/7B1M5 Llama052_ 4th image .. is that what you need?
- [00:02] <Llama052_> assuming the core switches connections are lagged?
- [00:02] <cnf> why are they all trunks?
- [00:02] <Llama052_> Do you really need all those vlans?
- [00:02] <Llama052_> but okay
- [00:03] <Llama052_> cnf, he doesn't have dvs iirc
- [00:04] <yancho> correct Llama052_
- [00:04] <Llama052_> so tag those LAGS onto vmware, but leave your vmware management vlan (whatever it may be) untagged
- [00:05] <cnf> right, time for bed
- [00:05] <cnf> gl
- [00:05] <Llama052_> add a vnetwork for the mgmt interface (with vlan specified) and one for everything else without a vlan defined
- [00:05] <Llama052_> cnf, halp
- [00:05] <yancho> thanks for ur help cnf .. gn :)
- [00:05] <Llama052_> and add all the nics to the vswitch for it
- [00:05] <Llama052_> assuming you have the LAG setup correctly on the switch it should be alive
- [00:06] <Llama052_> then setup untangle to look for those VLAN ids
- [00:06] <Llama052_> I'd put WAN and LAN on two different virtual nics though
- [00:06] <yancho> Llama052_: vnetwork is vlan?
- [00:07] <Llama052_> you'll want a kernel nic for the management
- [00:07] <Llama052_> whichever vlan that will live on
- [00:07] <Llama052_> it will need to know
- [00:07] <Llama052_> so you could do two vsiwtches, 1 for local and one for wan
- [00:07] <Llama052_> vswitches*
- [00:08] <Llama052_> select your vmkernal for management with a vlan, and then make a network for all of your LAN without vlan specified
- [00:08] <Llama052_> or Virtual machine port group
- [00:09] <Llama052_> you'll want to verify you have the vlans tagged on the switch connected to the vmware box
- [00:09] <Llama052_> vmware esx only needs to see the management vlan tbh, you can *untangle* the other vlans on untangle
- [00:12] <Llama052_> you've got about 5 more minutes before I head home for the night lol
- [00:12] <yancho> and management vlan has to be vlan 1 ? or can be any number?
- [00:12] <Llama052_> whatever you want to be able to access the vmware box on
- [00:12] <Llama052_> the management interface of the vmware box
- [00:12] <Llama052_> most people have a management vlan
- [00:12] <yancho> at the moment it is untagged .. unmanaged .. so we can do it managed 1
- [00:12] <Llama052_> dedicated for admin interfaces and stuff
- [00:13] <Llama052_> aslong as you can match it on the other side of the switch
- [00:13] <yancho> 1 on the switches is "default"
- [00:13] <yancho> so should be good
- [00:14] <Llama052_> alright then make a new port group for LAN
- [00:14] <Llama052_> and add all the LAN adapters to it, but don't define a vlan
- [00:14] <Llama052_> and do IP hash or Port ID load balancing
- [00:14] <yancho> lan adapters meaning the fibre nics correct?
- [00:14] <Llama052_> assuming the other side is setup as lag
- [00:14] <Llama052_> Whatever has all the local vlans tagged to it on the switch
- [00:14] <Llama052_> and is it a LAG on the switch
- [00:14] <yancho> we shall go for static lag I presume not lacp to play it safe right?
- [00:15] <Llama052_> LACP isn't supported without vds
- [00:15] <Llama052_> static
- [00:15] <yancho> oki :)
- [00:16] <yancho> seems like a plan! I guess I'll have a lot of meat to chew up tomorrow
- [00:17] <yancho> (well today lol .. its past midnight here :( )
- [00:18] <yancho> the 5 wans have to go each as a single port right? (due to ips etc)
- [00:19] <Llama052_> yeah you should be able to do that
- [00:19] <Llama052_> call the untangle folks since you purchased support, they can walk you through a lot of it
- [00:20] <yancho> and you're still of the opinion that i get an l3 before the l2 switches right?
- [00:21] <Llama052_> I'd want to get something to take layer-3 before your untangle device
- [00:21] <Llama052_> if you really expect 1000 concurrent end users
- [00:21] <Llama052_> especially with vlans
- [00:22] <Llama052_> all those requests need to go somewhere
- [00:22] <yancho> yes we can reach that numbers ... but most of the vlans are internet bound .. not much of local traffic
- [00:22] <Llama052_> vlan-to-vlan
- [00:22] <Llama052_> you can always see how it works as is
- [00:22] <Llama052_> also you have a single point of failure with the untangle box
- [00:22] <yancho> ok .. if we see untangle in pain .. we help with an l3
- [00:23] <Llama052_> you could do a vmware HA cluster, or get a backup untangle box
- [00:23] <Llama052_> I'd still research untangle bare metal if possible
- [00:23] <yancho> yeah i know that .. might be something for the future to get a backup untangle box .. on barebone with vrrm thingy
- [00:23] <Llama052_> goodluck
- [00:23] <Llama052_> I'm out
- [00:24] <yancho> vmware ha cluster is cool too .. still a problem with wans .. so might as well set up a mini untangle just in case
- [00:24] <yancho> many many many thanks for your help!!! thanks also to you Zew
- [00:24] <Llama052_> layer-3 switch would fix that
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement