Strange Cacti issue when talking to new host

[gwinans]

Howdy folks.

I have minimal control over the cacti install, but I'm trying to find where the actual problem exists.

Here's the situation: (Technical Support page is attached)

The Cacti instance itself is on a Windows server.

It has been able to connect without issue to an Amazon instance for the past year without issues.

We recently added a new instance (local, not at amazon). This instance of linux is a dedicated server running the same software versions as the Amazon instance. The net-snmp configs are also identical. For some reason, this new instance doesn't respond 100% of the time (more like 40% of the time) and causes connections to pile up until Cacti ... well, shits the bed.

There are no network issues between the Cacti server and the new machine. The only thing between them is a switch. No packet loss.

I'm at a loss as to why it works great on a remote instance (half way across the country) and is dying on an instance that's essentially local-net. Any thoughts? Suggestions?

[BSOD2600]

We recently added a new instance (local, not at amazon). This instance of linux is a dedicated server running the same software versions as the Amazon instance. The net-snmp configs are also identical. For some reason, this new instance doesn't respond 100% of the time (more like 40% of the time) and causes connections to pile up until Cacti ... well, shits the bed.

What do the logs show during the outage?
What uptime detection method are you using?

You do realize 0.8.8a is the latest, which has fixed a considerable number of bugs, possibly what you're experiencing too. http://www.cacti.net/changelog.php

[cigamit]

Also, what OS is the host that you are monitoring? What are your SNMP Timeout settings set to? How many data sources on that box?

I have this one Win 2008 R2 box that every query works fine on but Disk Usage. Just as soon as you run that, SNMP starts timing out on the box. I wonder if your issue is similar.