Hi!
This morning, I posted a spine issue, but since then, I realised it was probably more a poller.php bug.
https://github.com/Cacti/spine/issues/123
Hi!
Describe the bug
Just installed a brand new cacti. I added all my devices (1400+) and it is now polling.
It is configured to poll every 10seconds with poller.php called every minute by crontab.
/etc/cron.d/cacti
*/1 * * * * root php /usr/share/cacti-1.2.7/poller.php > /dev/null 2>&1

The problem is that instead of the 10seconds intervals, it happen in 8 seconds and sleep at the end of the 6 polling sequence. (see screenshot)
To Reproduce
Configure same parameter as I did I suppose...
Expected behavior
As you can see in my other screenshot of my current production server, the interval are 10seconds.
Screenshots

For comparison, here is the CPU stats from my other server 0.8.8g

Server (please complete the following information):
Cacti 1.2.7
Spine 1.2.7
RedHat 8.1
PHP 7.3.12
Compiling (please complete the following information):
Sounds like something on RHEL more than on spine since cron would be the mechanisim that would initiate spine
try this create a small script that echos the time to a text file and have it run at the same time restart the crond service to be sure it syncs up and compare if the echo and spine behave the same
/var/log/cron show that poller.php is launch by crontab exactly every minutes.
And as you can see in the screenshot, poller.php launch spine every... 8 seconds or so.
Man, 10 seconds with 1400 devices. Must be a big box. I would suggest that this is a small issue in poller.php.
Hmm, seems a counter issue inside of poller.php. It should record the start time, and the end time, take the difference between that and the start of the next poll, and then sleep for that amount of time.
Ok, that's what I tought.
FYI, running on a 20 x XeonE7-4850 v4 @ 2.10GHz
@+!
Yea, that's a beast.
Update your poller.php from branch 1.2.x and report back.
Yep, back to normal!

Thank you!