I’ve had a Dude implementation up for a couple of months, and have found a problem which has rendered it almost completely useless.
I want to monitor the state of remote sites which are semi-permanently connected by a PPTP link. The interfaces to use are therefore signified by (where X is the login name used).
If the link is configured to use the interface, everything works fine until the interface drops (the remote site logs out etc.). Then The Dude changes the interface for that link to a different interface, . The two picture bwlow shows the link definition before and after the interface drops.
The same applies to any interface which disappears and then reappears. Whilst this may not appear to be a problem, the fact that an interface has disappeared is important and should be reported to us. Consider the case of a remote site accidentally deleting a wireless link - The Dude will pick a different interface and then monitor that instead. We wouldn’t know about the interface going down until the end user contacts us. This picture below shows the state before and after a wireless link is deleted.
I can see no reason why The Dude should change the interface it’s graphing/monitoring/using - especially so since the change appears to be random and the link status will be incorrectly reported from that point on. Surely this completely defeats the whole point of monitoring software?
I have seen when I replace a piece of equipment the dude will just start monitoring what ever port on the new device that has the same index.
I agree with you that if an interface should go away the dude should not change but continue to stay on the same interface and complain that the interface disapeared.
It is possible to convert your links to standard links and then directly modify the label (appearance) to show the snmp information that you want. Then when the label goes away you know it has changed. I don’t have any virtual links that I need to track so I don’t have your trouble.
This would put a unix CPU load on a device label " CPU: [oid (“1.3.6.1.4.1.2620.1.6.7.2.4.0”)] ".
I don’t know if this would help at all but is there any way you can setup the PTP link so that it uses the same username and password every time?
I know the dude would still need to be fixed to stay on the same interface even if the above was the case. I am thinking you will need both for the dude to accurately track the ptp interfaces. Have to wait and see here.
Along those lines I have seen interfaces just forget what they used to be and have to go put them back but it has been a while since I saw that.
I have seen when I replace a piece of equipment the dude will just start monitoring what ever port on the new device that has the same index.
Hmmm. That’s not useful - if you have to check that everything’s set up properly before you can trust the results, you may as well do all the checks manually.
Perhaps the interfaces/ports should have a globally unique ID which is used instead of just the local index. Or, maybe The Dude should keep track of the text description of the interface/port and notify if that changes. Bit of a tricky one, but unfortunately, until I can trust that The Dude is monitoring what I set it up to monitor, it’s completely useless for me.
I don’t know if this would help at all but is there any way you can setup the PTP link so that it uses the same username and password every time?
Yes, I’ve set this up, so I want to monitor the interfaces , etc. The outbound connections all work AOK, but the inbound connections are the ones which disappear and reappear and cause the problems.
