I have Cacti 1.2.15 on Centos7, and keep running into an issue where all my graphs either stop collecting new data or come up blank - I presume new RRDAs are being created, as I see new ID numbers for the graphs showing up in the 'Graph List' view. I haven't tried digging under the covers too much yet, and I have found various posts that reference seeing this kind of thing happening, but not sure if a fix has been addressed yet. (Unfortunately, 1.2.16 and 1.2.17 don't appear on EPEL yet that I've seen, and I don't believe I see a fix listed in release notes that addresses this.)
Unfortunately, I'm not sure what I did that might have caused this. The first time around I edited an existing device or graph template, had the issue, and went ahead and deleted and recreated all devices.
This second time, however, all I think I did was add a new device. There was a week's gap between when I created the new device and when I next checked all the existing ones, so I'm not sure at what point it fell apart.
I have a feeling the old RRDAs are out there and being written to (the troubleshooter seems to support this) but the front-end is referencing brand-new, blank ones.
I might be able to tackle the SQL to fix or clean these up but at this point I don't really understand what's what... and hate for it to happen again.
Any suggestions on where to begin?
Thanks,
Eric
All graphs are blank, new rrds created?
Moderators: Developers, Moderators
Re: All graphs are blank, new rrds created?
For example, an SNMP bits in/out graph was at ID 313, and I still have an RRD file and the poller_item table still shows this.
But the graph_local table entry now shows ID 419 (no RRD file actually was created). No entry in graph_local for 313.
The poller (built-in, not spine) is still working and collecting data to old IDs/RRDs (for now, anyway... not sure what will happen if I restart it) but the front end is looking at new graph IDs that don't really exist.
Seems like this happened for all my SNMP bits in/out graphs...
Perhaps I can rename the RRDs to the new ID and update the poller_item lines? What else would have to get updated? Such a mess to try to do manually... and that's if I can figure out what was what.
But the graph_local table entry now shows ID 419 (no RRD file actually was created). No entry in graph_local for 313.
The poller (built-in, not spine) is still working and collecting data to old IDs/RRDs (for now, anyway... not sure what will happen if I restart it) but the front end is looking at new graph IDs that don't really exist.
Seems like this happened for all my SNMP bits in/out graphs...
Perhaps I can rename the RRDs to the new ID and update the poller_item lines? What else would have to get updated? Such a mess to try to do manually... and that's if I can figure out what was what.
Re: All graphs are blank, new rrds created?
If you delete and recreate your graphs, they get new ID's. As long as you didn't delete the data source then it can be reused.
Cacti Developer & Release Manager
The Cacti Group
Director
BV IT Solutions Ltd
+--------------------------------------------------------------------------+
Cacti Resources:
Cacti Website (including releases)
Cacti Issues
Cacti Development Releases
Cacti Development Documentation
The Cacti Group
Director
BV IT Solutions Ltd
+--------------------------------------------------------------------------+
Cacti Resources:
Cacti Website (including releases)
Cacti Issues
Cacti Development Releases
Cacti Development Documentation
Re: All graphs are blank, new rrds created?
Actually, I discovered that it was the change in polling frequency from 5 mins to 1 min. I reset it back to 5 and data started appearing again. I found other posts that hinted at this potential cause.
But what I don't understand is why did all historical data disappear from the RRD's? I could expect a lack of new data collection, but I wouldn't have expected everything to go blank.
Any insights on how this works?
Thanks,
Eric
But what I don't understand is why did all historical data disappear from the RRD's? I could expect a lack of new data collection, but I wouldn't have expected everything to go blank.
Any insights on how this works?
Thanks,
Eric
Who is online
Users browsing this forum: No registered users and 0 guests