DX NetOps

Hi All,

When I'm pinging a downed device, the expected return result would be "Request timed out", but it would be generated very slow.  It would take about 20 seconds for "Request timed out" message to appear.  Is this by design? 

In comparison, if I pinged a downed device from Windows command prompt, I would get "Request timed out" message every 5 seconds.

Is there anyway to make CA Spectrum console to return the message "Request timed out" back faster?

CA Spectrum v10.2.1

Thanks.

Hi,

CA Spectrum will provide response for device ping and poll status based on device responding to CA Spectrum one-click.

As you stated in your query that if you try to ping down device from windows CMD, it takes about 5 seconds to respond which means response from device is very slow. Anyways in order to debug delay issue kindly perform below steps.

1. Try to ping an active device and down device from same subnet of the down device for which you got delayed ping response from one-click console.

2. Do you see ping response time for down device also taking about 20 seconds? If yes, then network response time for that network subnet from Spectrum console is slow.

3. Note round trip time response ping for active device and if you see the ping response time is more for the active device then it is network response time for that network subnet from Spectrum console is slow.

Perform same ping process from one-click server CMD also and note the values.

Note: You can expect ping response time from One-click console is almost equal to 3x of round trip time since Spectrum performs ping check for down device multiple times to confirm.

Regards,

Vishnu Prasad

Hi Vishnu,

You misunderstood me.  I'm only concerned about devices that is currently down or not ping-able.

If I try to ping a bad device from Spectrum console I would get the "Request timed out" message, but it would take 20 seconds for each line.  I think this is too slow.

Regards,

Narongkij T.

Please check the DCM Timeout (default 3000 ms) and Retry Count (default 3) in the Modeling Information.

And VNM's Fault Isolation ICMP Timeout (3000 ms) and Retry Count (default 1).

You can see the detailed document here

Hi Naruhiro,

I have already checked those settings but does not help.  I'm still reading the document you linked, but can't find info that can possibly decrease my waiting time.

Regards,

Narongkij T.

What is ICMP Timeout and ICMP Try Count set to under VNM -->Fault Isolation in your environment? 

What you set here directly affect the timeout you see during ping 

It was originally 3 and I changed it to 1, but still the same.

What is ICMP Timeout set to (msec)?

Default at 3000, I changed to 1000 but still no effect.

I have checked on my lab server and I did not notice any slowness when a device is down from a ping from OneClick.

I suspect that there could be a performance problem with the SpectroSERVER as Notification and Timer latencies could cause this type of behavior.

I would suggest to open a support issue and we can check further with perfcollector9 data from your SpectroSERVER to assist you with possible performance problems.

Best regards,

Glenn

Was there ever a solution identified for this?  We are experiencing a similar issue running Spectrum 10.2.3.  If you ping a device from OneClick that is up, the response is as expected.  However, if the device is down, it takes about 20 seconds for each Request timed out to return to the ping window.

Thanks,

Michael

I believe all Spectrum is doing is a standard ping for the OS. Do you get the same timeout from the command line?

Joe

Joe,

If I ping directly from my Spectrum MLS server, the ping response is as expected.  It only seems to occur when using the OneClick GUI.

Michael

When I upgraded from v10.2.1 to v10.2.3 the problem went away.  But since you are at v10.2.3, have you tried restarting the service or restarting the server?  The newest version now out is v10.3 and we just updated to that to resolve some other issues, you may want to plan to update to the latest version.