On Wed, 23 Aug 2023, Corentin Labbe wrote:
Le Wed, Aug 23, 2023 at 03:56:27PM -0400, Josh Luthman a =E9crit :
For those that don't want to scroll through, 10k and 500k are the biggest but they are certainly outliers.
curl -s "
https://en.wikibooks.org/wiki/System_Monitoring_with_Xymon/User_Guide/The= _Xymon_Users_list"|grep Hosts|awk 'NF>1{print $NF}'|sort -V|tail 2247 2406 3062 3949 5339 5372 10940 569869
How many hosts are you looking to monitor?
For my own personnal case no more than hundreds.
But for my xython project, I want to know how far I need to bench to reach = (or for dreaming exceed) xymon performance.
But the problem for this list, it state only hosts and not number of tests = per hosts. This is not the same than just 1000 host with just ping test and 1000 hosts= with 10 network tests AND 20 "columns".
I've got about 2500 hosts ( in the xymon sense ). A large chunk are fake hosts for functionnal monitoring. A host ranges from 5 to 25 columns. Tests run every 5 minutes.
There is also a Vpage host that has ~2000 columns.
Load wise, the xymon server is on a VM with 8 vCPU and 8gb of RAM. Only 596Mb of RAM is actually counted as used by processes, and that would include apache and everything else.
The only thing that has been clunky is the multi-server failover, but it works well enough.