HPE 5130 dropping pings

[mhavu]

I have a workstation and a bunch of embedded real-time systems in a private network. (I would attach an image, but I guess that is not allowed for new users.) The number of embedded devices increased, so I replaced our HP Procurve 2524 switch with a new HPE 5130-48G-PoE+-4SFP+. The workstation has IP address 192.168.1.1, and the embedded devices span a continuous range of IP addresses in the 192.168.1.0 subnet. Bridge configuration is fine for this purpose (except that I would like to be able to remotely manage the switch from the workstation), so I started testing with the switch factory settings.

Pinging the embedded devices from the workstation with increasing packet size caused lost packets already at little over 8 kB. I enabled jumbo frames and burst mode, but these didn't have any effect. Does anyone have any ideas what might be causing this? I also seem to get ten to hundred times longer latencies in the private network than I do with the same workstation in a public network with an identical switch.

Another question: After the tests, I enabled remote management. I can now use the management GUI, but I would rather use CLI. However, I get "Connection to 192.168.1.2 closed by remote host" as soon as SSH accepts the password. Any hints on what might still be missing from the configuration?

[Colin_Painter]

Hi,

Are the following commands added on to the device? I've had issues with ping/traceroutes before an using these helped.

ip unreachables enable
ip ttl-expires enable

[sdide]

Hi,

I'll try to address some of your points.

Pinging the embedded devices from the workstation with increasing packet size caused lost packets already at little over 8 kB.

Was this also an issue with the Procurve-switch, or is this a newly arisen problem (when you deployed the 5130)? Embedded devices are usualy small with close to very little CPU. Sending large frames to the CPU might not just be a good idea.

I also seem to get ten to hundred times longer latencies in the private network than I do with the same workstation in a public network with an identical switch.

What do mean here? Public Network vs private network. Do you mean on a network using public addressing (i.e not RFC 1918 addresses?)  Anyway. Increasing ping times when pinging "through" the switch is not related to the switch. (Switch just forwards packets)

Another question: After the tests, I enabled remote management. I can now use the management GUI, but I would rather use CLI. However, I get "Connection to 192.168.1.2 closed by remote host" as soon as SSH accepts the password. Any hints on what might still be missing from the configuration?

You need to do a few things before ssh works.

public-key local create rsa
line vty 0 63 
 authentication-mode scheme 
 protocol inbound ssh
 quit
domain <name-of-your-domain>
 <define-your-aaa-methods> # tacacs+ / radius / local
 quit
default domain <name-of-your-domain> enable

Also, if your method is local, you want to create a local user for that purpose. 

local-user <local-user-name> class manage
 password hash <password-hash>
 service-type ssh
 authorization-attribute user-role network-admin #user can do anything
 quit

Regards