Cacti 1.2.26 Poller Output Table not Empty with only 2 devices defined

[Rno]

Hello,
I have a strange issue with my new server (again)!
I'm using spine on cacti 1.2.26 php 8.1.14.
I have a cron job every minute:

Code: Select all

*/1 * * * *    apache  /usr/bin/php /usr/share/cacti/poller.php > /dev/null 2>/DATA/cacti-log/poller-error.log

On Cacti the poller config is:
Poller interval evry minute
Cron/Daemon every minute.

Spine has the correct right:

Code: Select all

ls -l $CACTI/spine/bin/spine
-rwsr-sr-x. 1 root root 659128 Jan  3 10:14 /usr/share/cacti//spine/bin/spine

The system stats is the following:

Code: Select all

05/01/2024 08:23:03 - SYSTEM STATS: Time:2.3137 Method:spine Processes:5 Threads:12 Hosts:2 HostsPerProcess:1 DataSources:39 RRDsProcessed:0

The number of process and thread is based on the config the server would need when everything is in place, so far it's too big, but I don't think it's the issue.

Running spine on the CLI with:

Code: Select all

 sudo -u apache ./spine -R -V 5

No error at all, icmp ping is ok

But doing so from cacti give me a icmp ping timeout (if it's related to my problem)
And device is still up, even if Downed Device Detection is setup as PIN and SNMP Uptime

But still every 5 minutes I got an error telling me that WARNING: Poller Output Table not Empty

Any clue where to look ?