cpu & memory probes sometimes (more than desired) give false positives.


J. Boardman

Hi ,


What is the hardware you are use .

The dude is running on a dedicated Dell Poweredge 650 server / win XP.

The monitored gear we are having issues with those two probes are routerOS different hardware, X86 & Routerboards


Regards

mmmmmmmmmmmm



Go to control panal >>> System >>> advanced >>> performense >>> Then choose adjust the best performense


Try It

mmmmmmmmmmmm



Go to control panal >>> System >>> advanced >>> performense >>> Then choose adjust the best performense


Try It

Hi, thanks for your reply, that's the way it is, and the way it was ... any other idea ?

i see that when the link between the dude server and the "monitored device" is bad, are these remote or local devices?

Local and Remotes, no difference.

We had similar sort of problems when we were monitoring 300-400 devices with a single Dude machine (Win XP - I think). We tried to figure out the problem and couldn't pin it down to CPU, Memory or network throughput. It was a problem that was very intermittent and we could not pin-point what the cause of it was. In the end we figured that it may be better to have multiple Dude agents, each polling their own subset of devices so as to distribute the processing. We ended up building this configuration putting six Dude machines (Win2K3) as VMs on an ESX server. Each Dude machine has a separate physical NIC connected to a different network segment and they all communicate with each other through the ESX "Host Only" network. One Dude server is the master that all clients connect to and all the other Dude servers are agents.

At the same time as we changed to this topology we also went from Version 3.something to 4.0 beta 2 so the fact that everything has been stable and working reliably for many months is not easily attributable to any one thing. There are now 658 items in the Device table.

We had similar sort of problems when we were monitoring 300-400 devices with a single Dude machine (Win XP - I think). We tried to figure out the problem and couldn't pin it down to CPU, Memory or network throughput. It was a problem that was very intermittent and we could not pin-point what the cause of it was. In the end we figured that it may be better to have multiple Dude agents, each polling their own subset of devices so as to distribute the processing. We ended up building this configuration putting six Dude machines (Win2K3) as VMs on an ESX server. Each Dude machine has a separate physical NIC connected to a different network segment and they all communicate with each other through the ESX "Host Only" network. One Dude server is the master that all clients connect to and all the other Dude servers are agents.

At the same time as we changed to this topology we also went from Version 3.something to 4.0 beta 2 so the fact that everything has been stable and working reliably for many months is not easily attributable to any one thing. There are now 658 items in the Device table.

Thanks for sharing ...