Odd load cycles.
I'm running a linux cluster and I use MRTG to graph the system load, and I'm noticing an odd load cycle.
The two systems affected are the DB server (only running mysql) and the webserver (only running Apache).
In both cases the load peaks at .6 and the trough is 0.00.
The load cycles are almost perfect sine waves. The odd thing is the frequency of the wave cycle is not 24 hours. The db server goes through exactly 6 cycles over 7 days and the webserver is exactly 5 cycles over 7 days. The beginning/end of the week is always on a trough for the web server and the db server always begins/ends the week on the crest of the wave.
If this load was cron based, I would expect a 24 hour cycle. And if it was user based I would expect a 24 hour cycle with a lull on weekends. This is not the case.
There seem to be no unusual proceeses running during the higher load periods.
While this seems to be doing no harm I am very curious as to what is causing this oddly load cycle.
Any hints?
These systems are running linux kernel 2.4.20-31.9. It's the RedHat 9 distro.
The two systems affected are the DB server (only running mysql) and the webserver (only running Apache).
In both cases the load peaks at .6 and the trough is 0.00.
The load cycles are almost perfect sine waves. The odd thing is the frequency of the wave cycle is not 24 hours. The db server goes through exactly 6 cycles over 7 days and the webserver is exactly 5 cycles over 7 days. The beginning/end of the week is always on a trough for the web server and the db server always begins/ends the week on the crest of the wave.
If this load was cron based, I would expect a 24 hour cycle. And if it was user based I would expect a 24 hour cycle with a lull on weekends. This is not the case.
There seem to be no unusual proceeses running during the higher load periods.
While this seems to be doing no harm I am very curious as to what is causing this oddly load cycle.
Any hints?
These systems are running linux kernel 2.4.20-31.9. It's the RedHat 9 distro.
