Posted on July 2, 2015 3:18 pm
 |  Asked by Jeff
Print Friendly, PDF & Email
Marked as spam
Posted by Philippe Bureau
Answered on July 2, 2015 3:48 pm


Would you explain further what you are trying to do?

You can use the ”copy” command to get a file from a server.

You must have an IP address configured on the management interface and reachability to the server first.

If you want automated image update when a switch is added to your network through the management interface, you can use ZTP server.  See :



Posted by Alexis Dacquay
Answered on July 2, 2015 4:26 pm


The way you transfer an EOS software image to an Arista switch is not different via the Management interface than it is with in-band management.

You can ssh to the Arista switch with the out of band interface IP (interface Management 1) or in-band interface (for example on an SVI or a routed port).

Once you are logged on the switch, you can use the ”copy”. You would find more details about the ”copy” command in the manual

Arista(config)#copy ?
  boot-extensions       Copy boot extensions configuration
  certificate:          Source file path
  clean-config          Copy from clean, default, configuration
  drive:                Source file path
  extension:            Source file path
  file:                 Source file path
  flash:                Source file path
  ftp:                  Source file path
  http:                 Source file path
  https:                Source file path
  installed-extensions  Copy installed extensions status
  running-config        Copy from current system configuration
  scp:                  Source file path
  sftp:                 Source file path
  startup-config        Copy from startup configuration
  system:               Source file path
  terminal:             Source file path
  tftp:                 Source file path

For example:

Arista#copy scp://user@server/path/EOS-4.15.0F.swi flash:
user@server's password:
EOS-4.15.0F.swi                                               100%  348MB  21.8MB/s   00:16





Marked as spam

This did not answer his original issue. Now I am having the same issue. I cannot copy scp to the remote server once the mgmt0/1 is located in VRF MGMT….. Any work arounds?

(Shawn at February 9, 2016 4:10 pm)

Hi Shawn,

This is a guess on our part but most likely you have to switch your routing-context to the VRF and then issue the SCP. Once you put the Management interface in a VRF it is in a separate routing table. If you are issuing the SCP from the default vrf you will not utilize the management interface ip address as the source ip and most likely, based on your connectivity and policies, that is going to cause issues as you are utilizing the interfaces/ip’s within the default vrf to access your management servers which is probably not allowed.

In order to issue the scp from the management vrf please issue the following command "routing-context vrf vrfname". Please note this is a privileged exec mode command. Also, if you haven’t changed your prompt then once you issue the above command your prompt will change to reflect the vrf. Once you have switched to your vrf routing context then issue the scp from there. HTH

If this does not solve your issue then could we please request that you provide the following:

a) Your entire configuration in a text file

b) A screenshot and/or Cli output of you issuing the command in question.

c) Please advise if you are able to reach your Mgmt default gateway and if the host you are trying to reach is within the same subnet or outside the subnet. Include ping Cli output if possible.

d) If the host is outside your Mgmt subnet then do you have a route defined within the VRF to reach that host.

Below is an example for your reference. HTH

localhost#sh vrf Maximum number of vrfs allowed: 15 Vrf RD Protocols State Interfaces ----------- ------------- --------------- -------------------- ----------- Mgmt 1:1 ipv4,ipv6 v4:no routing, Management1 v6:no routing localhost#routing-context vrf Mgmt localhost(vrf:Mgmt)#

(Vikram at February 9, 2016 6:32 pm)
Posted by Jeff
Answered on July 6, 2015 3:27 pm

The problem wasn’t the management interface at all. The problem was actually that we had put that interface into a vrf and I couldn’t route out of that vrf. Once I moved that interface out of the vrf, all was well. Thanks for all the suggestions and help.

Post your Answer

You must be logged in to post an answer.