Migration from 0.8.8b to 1.2.25, about HALF of graphs not updating with new data

[joseph85750]

A few weeks ago I migrated from 0.8.8b to 1.2.25, following this procedure:

http://forums.cacti.net/viewtopic.php?t=59586

For the most part I got everything working, but only about half of my rrd files are getting updated.
I'm wondering if it has something to do with the procedure for migrating the rdd files from the old system to the new.
For example, in the procedure above, I saw this process which involved a rrdtool dump/restore, which is what I did:

http://forums.cacti.net/viewtopic.php?p=276851#p276851

But later I saw a thread where the files are just copied from old to new server as is:

http://forums.cacti.net/viewtopic.php?p=279004#p279004

Could there be some version mismatch?

[macan]

Check rrd files permissions
Check cacti log

You can try Troubleshooting->Data sources

[joseph85750]

All the RRD files have the correct permissions. They were all imported from the old install at the same time.
As a test, I tried to directly copy one of the working RRD files from the old functioning server directly (no conversion) to the new system where that same file seems to have an issue, but upon doing that an error image was displayed upon viewing that particular graph:

cactiProblem.png (41.32 KiB) Viewed 1642 times

I restored the RRD file, and the error went away. But I was again left with an image with no data.

Also, the RRD files in question are not getting any updates. The timestamps are not changing.

This is what I see in my logs:

Code: Select all

2023-12-05 13:24:01 - CMDPHP WARNING: cmd.php poller has run over its polling interval and therefore is ending
2023-12-05 13:24:00 - POLLER: Poller[Main Poller] PID[2421942] WARNING: There are 1 processes detected as overrunning a polling cycle, please investigate 

[TheWitness]

Move to spine from cmd.php and increase both processes and threads to get your polling time down to reasonable measures.

[joseph85750]

Moving to spine solved my problem, but I still needed to use 5-minute intervals instead of 1-minute, since it took about 2 minutes to complete a run on a dual core/cpu system.