Y
yvette.ye@gmail.com
I have been monitoring a server with 4 G RAM, 4 CPUs, the disk: is at
a 14 SAN RAID 5 disk group, I have setup the alerts, the following
are some of the performance counters show in Event Log, I have some
questions about them:
1) Counter: \\servername\Process(msmdsrv)\% Processor Time has
tripped its alert threshold. The counter value of 792.187223329792 is
over the limit value of 99.
my question is: How is possible a CPU% is over 100%? Is it because
there are 4 CPUs and it shows as 8 CPU instants in Windows, and number
792 means 792% when it sums up for 8 CPUs?
2) Counter: \\servername\System\Processor Queue Length has tripped its
alert threshold. The counter value of 15 is over the limit value of
8.
Does that means the CPU is very busy? the busy level for processor
Queue Length is 2, than 4 CPU is 8. but does it support to be 16, as
there are 8 CPU instants in windows.
3) Counter: \\servername\Memory\Pages/sec has tripped its alert
threshold. The counter value of 173.048969603907 is over the limit
value of 20.
Does that mean the memory is extremely busy? compare 20, the 173 is
extremely extremely high, is it 20 a good base number to set?
4) Counter: \\servername\PhysicalDisk(_Total)\% Disk Time has tripped
its alert threshold. The counter value of 176.393046730998 is over the
limit value of 48.
Does that mean the disk is extremely busy? since there are 24 drives,
24 * 2 disk queue length = 48. but 176 is well too high still.
There is a performance issue on this server, and what do you think
could be the cause? and do you think those measures are a good set for
monitoring?
How can I setup the perfermon monitor to only trigger the alert when
the counter is consistent high over a period of time, intead of
trigger it right away even there is a spike. I don't think count a
spike is a good idea.
Thanks,
fshguo.
a 14 SAN RAID 5 disk group, I have setup the alerts, the following
are some of the performance counters show in Event Log, I have some
questions about them:
1) Counter: \\servername\Process(msmdsrv)\% Processor Time has
tripped its alert threshold. The counter value of 792.187223329792 is
over the limit value of 99.
my question is: How is possible a CPU% is over 100%? Is it because
there are 4 CPUs and it shows as 8 CPU instants in Windows, and number
792 means 792% when it sums up for 8 CPUs?
2) Counter: \\servername\System\Processor Queue Length has tripped its
alert threshold. The counter value of 15 is over the limit value of
8.
Does that means the CPU is very busy? the busy level for processor
Queue Length is 2, than 4 CPU is 8. but does it support to be 16, as
there are 8 CPU instants in windows.
3) Counter: \\servername\Memory\Pages/sec has tripped its alert
threshold. The counter value of 173.048969603907 is over the limit
value of 20.
Does that mean the memory is extremely busy? compare 20, the 173 is
extremely extremely high, is it 20 a good base number to set?
4) Counter: \\servername\PhysicalDisk(_Total)\% Disk Time has tripped
its alert threshold. The counter value of 176.393046730998 is over the
limit value of 48.
Does that mean the disk is extremely busy? since there are 24 drives,
24 * 2 disk queue length = 48. but 176 is well too high still.
There is a performance issue on this server, and what do you think
could be the cause? and do you think those measures are a good set for
monitoring?
How can I setup the perfermon monitor to only trigger the alert when
the counter is consistent high over a period of time, intead of
trigger it right away even there is a spike. I don't think count a
spike is a good idea.
Thanks,
fshguo.