a screenshot would tell me a lot more info than what you summarized. what graph for example… Maybe other info you don’t see.
Assuming you mean the ports graph. There could be an issue with partial data fetching. The the ifTable walk fails mid walk, it could remove ports. But the ports are only soft deleted so they should reconnect when next polled successfully. Unless someone is clicking the port purge button in that time. That would lose graphs.
Nobody is clicking anything, I am the only user of the system. The graph data doesn’t return until I reselect the port. I can provide a screenshot, but it will be a white box that shows the txt message of what I already typed above, if you really want that I can get one, but I don’t see how that will help you.
Anyway, I think we are going in two different directions. I need to revisit your recommendation to change it to port association. If that doesn’t work, then I’ll probably move back to PRTG, this never happened when I used PRTG.
What does this mean? Can you give a little more detail?
I wouldn’t push so much but I really take data loss seriously and would like to know what is happening. As I said this is the only report.
I’m not explaining it again. I’m not sure which part you don’t understand, you can go back and read again if you’d like. The device is off during a scan and it forgets the port, it is that simple. I changed 1 device to what you recommended previously, I’ll see if that device breaks like the other devices.
I’m assuming you mean a dashboard graph widget that had a port selected. If it now shows a stale port ID reference, that suggests the port lost its association, likely due to an ifIndex change on the device.
Setting the port association mode to ifName should prevent this and resolve the issue going forward.
Yes, a dashboard graph the widget it port bits and yes it shows a stale port.
A few days ago I did change one to ifName, hopefully you are right and that solves the issue. I did see there is a way to change it for all devices and make it the new default but I just want to test with 1 device for now.
I’m not sure why ifName is not the default, this is an extremely annoying issue and it sounds like it is also the problem for the new RRD graphs or the loss of the data for my graphs. I really liked having that historical data.
Also, this is not obvious, I’m sure, but the Traffic Bills also loses the port (which makes sense to me now that I know the problem) so in addition to having data lost on the graphs, any traffic bill is also broken/has no data. I don’t actually bill anyone, I just like how granular the Traffic Bills are (day, hour, month, etc…) which is why I use it.
Because even more devices have unstable ifName than devices that have unstable ifIndex.
Now this sounds like a LibreNMS problem. Which is it? ifName is also not good?
Why is LibreNMS dropping devices but the two other SNMP programs I used never did this? The same devices were being monitored in all the programs I was testing.
At first it seemed like there was a solution, switching to ifName, now it sounds like that is also not reliable.
It is a device problem. The problem is there was no snmp validation out there and a lot of vendors implemented snmp poorly.
Here is what the RFC says about ifIndex.
"A unique value for each interface. Its value ranges between 1 and the value of ifNumber. The value for each interface must remain constant at least from one re-initialization of the entity's network management system to the next re initialization."
The main offender is vpn interfaces because the network interface can be transient. Being added and removed. Now, ideally they would have the same ifIndex when they come back, but some vendors give them a new ifIndex every time.
As for ifName it makes no stability claims. But one of the biggest problems isn’t it changing, but simply being blank for every interface. Although, Mikrorik for example allows users to set interface names. (Normally user supplied names/descriptions go in the ifAlias field)
If you can come up with a solution that works 100% correctly all of the time, is send you a case of whatever your preferred beverage is. 
Oh and sorry for your specific problem, the solution should be switching to ifName.
The problem for LibreNMS is we cannot set ifName generically as it won’t work for every single device in every single situation.
I disagree that it is a device problem based on the data I have. I’ve never experienced this with PRTG. PRTG just shows a gap of missing data from when the device was offline. LibreNMS also does this but it is hit or miss when the gap is there vs when the interface is removed from graphs and billing, etc.
I did notate the one device I’ve switched to ifName hopefully that solves the issue. Time will tell.
It seems they have the exact same problem.