The wait operation timed out (Performance Counter error 0x102)

What operating systems are you using (on your Exchange and the PRTG server)?

Server 2008 R2 Enterprise Edition for Exchange 2010. Server 2008 R2 for the PRTG server.

Strange - have you tried increasing the interval for those specific counters? Additionally, you can use any character as unit - it should resemble the value however :)

So just to clarify - the Units value is purely a label and has no bearing on if the data is displayed correctly?

I've recreated that counter and set it to check every 5 minutes rather than our default of every 60 seconds, and I still get the same error.

I can read the counter successfully from the Perfmon mmc with checks in any interval I like from 1 per second upwards.

Yep, purely a label. Could you try to install a remote probe on the server to see if it runs more stable then?

I'd suggest the KB page https://www.paessler.com/manuals/prtg/perfcounter_custom_sensor be updated to make that clear. I've been tearing my hair out trying to find what I need to set Units to.

Adding a separate probe seems excessive for a single perfmon counter. I've also tried creating a new custom sensor copying and pasting the values from the example for a different machine and get the same result.

Interestingly I also get the same error if I try to add the "PerfCounter IIS Application Pool" sensor. All our other Perfmon checks seem to be coming in via WMI.

Have you seen this thread already? Especially everything performance counter related, starting with "The re-enable performance counters option".

That article seems to be talking about fixing things on the server being monitored, however the counters work fine through PerfMon both locally and remotely so I'm not sure how it's relevant.

Sorry for the delay in publishing/responding. What happens when you actually pause and resume the sensor when it's in error state? Does it work properly then?

Pausing then resuming makes no difference. I get the same result on a couple of different checks to separate servers from the same probe.

I've just created a new check from a different probe to double check things and that one is mostly working, although every third entry in an error - it's consistently recording two values then erroring, then recording another two values and erroring again. If I check the event log in PRTG it shows "The wait operation timed out (Performance Counter error 0x102)" here too as the error condition when it's failing.

Any update on this? Restarting the probe services occasionally gives us a short period of successful monitoring, but it always fails again before long.

Did you already consider deploying a Remote Probe directly onto the target server? Then the monitoring would be local via localhost which usually is much less troublesome.

Hello, restart of PRTGProbeService on the PRTG server solve the problem. But it is workaround.