Package: cacti Version: 0.8.6c-8 Followup-For: Bug #309786 Hi,
I've come across this issue too and manually set the maximum CPU to 200% or 400% (or whatever's appropriate for the machine being graphed). It'd be nice if cacti could set this automatically. I think this information can be obtained from an SNMP query, on a Linux machine there's these on a single HyperThreading P4 machine: HOST-RESOURCES-MIB::hrDeviceType.768 = OID: HOST-RESOURCES-TYPES::hrDeviceProcessor HOST-RESOURCES-MIB::hrDeviceType.769 = OID: HOST-RESOURCES-TYPES::hrDeviceProcessor Anyway that'd be extra programming that I guess is a wishlist/upstream issue. However, after I manually set the limit to 100 * (number of processors) it seems that any individual value over 100 gets ignored. For example in the attached image the bits where there's no system CPU usage shown (in red) is when system CPU usage was over 100%. This is a dual HyperThreading Xeon system and the CPU limit was manually set to 400. (The reason for the large amount of system CPU usage is a degraded RAID array which seems to be causing slower disk I/O system calls and also occasionally some processes to go a bit mad and sit there chewing up CPU). So it looks like there's still a 100 limit being enforced somewhere beyond just the settable maximum graph value. I don't know if this is in snmpd or cacti though. Cheers, Mark. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
