PLA_E_DCS_ALREADY_EXISTS (0x803000B7) quick fix

The 30-second fix: delete or rename the collector set

This error pops up when you try to start a Data Collector Set in Performance Monitor (perfmon.msc) and Windows tells you the thing already exists. Common scenario: you're setting up a custom performance log for a client's SQL server, hit start, and get slapped with 0x803000B7. Don't panic.

Open Performance Monitor (Win+R, type perfmon.msc). Expand Data Collector Sets > User Defined. Find the set you just created. Right-click it and choose Delete. Yes, it's that dumb. If you want to keep the configuration, right-click and Rename it instead. The error means the name you picked is already taken somewhere in the system. Had a client last month whose print queue died because of this exact same conflict — the name collided with a shadow copy set.

If you don't see it in User Defined, check Event Trace Sessions and Startup Event Trace Sessions. Sometimes they hide there. Delete the duplicate, then re-run your set. Takes 30 seconds.

5-minute fix: clear the registry cache

If the simple delete didn't work, the system's registry cache is holding onto a ghost. Here's the fix:

Open regedit as Administrator.
Navigate to HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Perflib
Under Perflib, look for a key named Last Counter and Last Help. Don't touch those — they're for performance counters.
Instead, go to HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\PerfProc\Performance
Delete any value named Disable Performance Counters if it's set to 1. That's a common cause of ghost sets.
Reboot.

I've seen this happen after a Windows update botches the performance counter registration. Real example: a client's Windows 10 22H2 machine got this after a cumulative update. Clearing that registry key fixed it.

15-minute fix: rebuild performance counters entirely

If you're still stuck, the entire performance counter database is corrupted. Skip trying to fix individual items — just rebuild them. Open an elevated command prompt (Win+X, choose Terminal Admin) and run these commands in order:

lodctr /R
cd \windows\system32
lodctr /E:PerfProc
lodctr /E:PerfDisk
lodctr /E:PerfNet
lodctr /E:PerfOS

The lodctr /R command rebuilds the counter list from scratch. It's like a factory reset for performance counters. Takes about 10 minutes to complete fully. After it finishes, reboot and try your Data Collector Set again.

If that fails, you might need to manually register the counters:

cd c:\windows\system32
regsvr32 /s perfproc.dll
regsvr32 /s perfos.dll
regsvr32 /s perfnet.dll
regsvr32 /s perfdisk.dll

Had a client whose entire print queue died because of this — turns out a corrupted performance counter broke the print spooler's monitoring component. The rebuild fixed both issues.

When to call in backup

If none of this works, your Windows installation might be damaged beyond quick repair. Run sfc /scannow and DISM /Online /Cleanup-Image /RestoreHealth. I've only seen this once in 15 years — a server that had been patched with a third-party tool that nuked the performance subsystem. Had to restore from backup.

But honestly, 98% of the time the delete or rename fix works. Don't overcomplicate it.