We’ve noticed that when an application isn’t responding (as is the case with the bug on powerdns.py), LibreNMS just seems to go and assume that it will not return and deletes the data entirely.
This should not happen without some safeguards in place. When that application does come back, the data is reset to new, losing the history.
There needs to be an improvement on this either that the data is kept for a period of time so if it does resume, it just adds to the original data or deletion only after a reasonable or defined period (i.e 30 days).
Maybe u got someting in your config like $config['rrd_purge'] ?
If so… what is is it set to?
Naw, he is right, several things do this. They probably need to be soft deleted so they can be restored.
We don’t delete our rrds automatically.
On looking further, they actually remain on disk but with different numeric IDs, so you can copy back over the newly created one but in any case Libre seems to be a bit trigger happy with it and needs to give it more time before entirely removing the link.
Or at least, re-using an old RRD when the application comes back if the data is still there to use.
Yust mind asking, for how long wass the polling for this app not responding be4 it reappeerd? (As an new id)
Thanks in advance.
Oh thats long,
Did it reappeer at an discovery run?
Yust thinking…
So 1st discovery “remove” from polling, last Discovery whenn it whent (alrady) back online, u got an new ID?
I assume that is what happened.
In any case, that’s not really a long time but I would still say that regardless, it should use the same RRD as the source of the data is the same.
For interfaces, this seems to happen if it stops and then comes back…