Switching

Hello,

i have 2x qfx3500 i want to know can i use 4x 10G ports for virtual chassis and not qsfp port?

thnaks,

Hi,

As per the documentation you should be able to use the 10g ports.

The following ports on QFX Series switches that support Virtual Chassis can be configured into VCPs to form a QFX Series Virtual Chassis:

• On QFX5200 switches: Any 40-Gigabit Ethernet QSFP+ ports

   

  NOTE

  Starting in Junos OS Release 17.3R2-S4, 100-Gigabit Ethernet QSFP28 ports are also supported as VCPs on QFX5200 switches.

• On QFX5120-48Y or EX4650-48Y switches: Only the eight 40-Gigabit Ethernet QSFP+ or 100-Gigabit Ethernet QSFP28 front panel uplink ports (ports 48 through 55)

  On QFX5120-48T switches: Only the six 40-Gigabit Ethernet QSFP+ or 100-Gigabit Ethernet QSFP28 front panel uplink ports (ports 48 through 53)

• On QFX5110 switches: Any 40-Gigabit Ethernet QSFP+ or 100-Gigabit Ethernet QSFP28 ports

• On QFX3500, QFX3600, or QFX5100 switches: Any non-channelized 40-Gigabit Ethernet QSFP+ ports

• Any fixed 10-Gigabit Ethernet SFP+ ports on any QFX Series switches that support these ports

Hope this helps

and I can add that the 4 x 10G will automatically be joined in a LAG between the devices:

You can increase VCP bandwidth between member switches by connecting multiple VCP links between the switches. When multiple VCP links interconnect the same two member switches, the links automatically form a Link Aggregation Group (LAG) bundle if they have the same speeds. For example, if you have two 40-Gigabit Ethernet QSFP+ interfaces configured as VCPs between member switches, the two links form a LAG with 80-Gbps of total bandwidth.

You can see this after connecting the ports by looking at "Trunk ID" as seen below where the 2 x 40G are bundled in a 80G lag:

user@ex4600-vc> show virtual-chassis vc-port
fpc0:
--------------------------------------------------------------------------
Interface   Type              Trunk  Status       Speed        Neighbor
or                             ID                 (mbps)       ID  Interface
PIC / Port
0/24        Configured          5    Up           40000        1   vcp-255/0/24
0/25        Configured          5    Up           40000        1   vcp-255/0/25

fpc1:
--------------------------------------------------------------------------
Interface   Type              Trunk  Status       Speed        Neighbor
or                             ID                 (mbps)       ID  Interface
PIC / Port
0/24        Configured          5    Up           40000        0   vcp-255/0/24
0/25        Configured          5    Up           40000        0   vcp-255/0/25

{master:1}
user@ex4600-vc>

1) so if i connect port 44-47 from qfx 3500 #1 to port 44-47 qfx 3500 #2 and follow http://tawfique19.blogspot.com/2017/08/configuring-qfx5100-two-members-virtual.html i can have 4x 10 as AE (LAG) and 40G throughput between my switches?

2) in http://tawfique19.blogspot.com/2017/08/configuring-qfx5100-two-members-virtual.html there no explaination that where should i set ports for VCP , would you guide me as well ?

because i think i should set somewhere that for example port 44-47 are using for my VCP, or junops detect this automatically?

Yes, you will have 40G between your QFX3500's... but only up to 10G per flow due to hashing across the 4x10G links.

To activate ports as VC ports, you do something like this on both nodes.

request virtual-chassis vc-port set pic-slot 0 port 44
request virtual-chassis vc-port set pic-slot 0 port 45
request virtual-chassis vc-port set pic-slot 0 port 46
request virtual-chassis vc-port set pic-slot 0 port 47

Thank you,

1) and would you guide me with a set of commands for configure 2x qfx3500 as virtual chassis? because i read manya rticles but i want to choose best one,

2) when i have configured 2x qfx3500 as vc one of my top of rack switch 10g port will connect to qfx3500 port1 and my second 10g for my top track switch will connect to qfx3500 #2 , i want to know how can i manage ports for backup device in VCP ? (i want to configure AE/LACP for my top of rack 2x 10g)

THanks,

1)

I would do it preprovisioned - just add the right serial numbers:

set virtual-chassis preprovisioned
set virtual-chassis no-split-detection
set virtual-chassis member 0 role routing-engine
set virtual-chassis member 0 serial-number XXX
set virtual-chassis member 1 role routing-engine
set virtual-chassis member 1 serial-number YYY

2)

When the two switches are joined in a VC, then you configure a LAG just like a standalone switch, only difference is interface numbering; member0 interfaces will start with eg. xe-0 and member1 interfaces start with xe-1.

Example config for a LAG on the first two physical ports in the QFX3500s with LACP enabled doing an access port in vlan XXX:

set chassis aggregated-devices ethernet device-count 10
set interfaces xe-0/0/0 ether-options ae0
set interfaces xe-1/0/0 ether-options ae0
set interfaces ae0 aggregated-ether-options lacp active
set interfaces ae0.0 family ethernet-switching vlan members XXX

Jonas, thank you so much for your help,

last question is when i set ip address on management interface port for master switch then there is no need to set management ip for second witchswitch? and in failure cases when backup switch elect as master then use same management interface ip ?

to ensure the management IP is being moved to the master routing-engine, please configure the IP on vme.0 like this:

set interfaces vme unit 0 family inet address 10.0.0.10/24

@blackmetal vme is a virtual management IP address that follows the Master switch.  Master switch could move between members, but the VME IP will always be responded to by current Master.  It is only on the Master that most config commands are applicable.  Once connected to the Master, you can move to other members within the VC via:

request session member <member-id>    you should also check this KB below for just in case.  This KB also shows you standard output for above command.  This could be used for specific troubleshooting, etc. but not config.

https://kb.juniper.net/InfoCenter/index?page=content&id=KB16348&cat=SWITCHING&actp=LIST

You could also assign individual IPs to each member of a VC via using their me.0 interface.  VME is the recommended way to go.

if i keep backup switch untouch and configure managemnt ip only on master switch , when master switch fail then backup switch elect as master device, does the managemnet ip move to backup?

Using vme interface yes.  Using only me, as me for each member would need to be unique.  Config is only allowed via the current Master.

With vme you need all physical me ports (that could be the Master RE) connected to the same physical network (vlan/subnet).  This is pure OOB.

For inband, any IRB IP address is associated with Master RE/CPU from a Management perspective.  The IRB functions (routes in hardware) for all members.

Again, if needed, once you are on any member (console or remote), you can move to any other member.  Only Master member allows config, which applies to whole VC.

ok got it,

so when i connect all qfx mgmt port to same network and set ip address on vme all members can use that ip address when elect as master, am i right?