I've been trying to figure out the cause of some unexpected CPU and I/O spikes on our print server, which is running on Windows 2022. Every 5 minutes, it seems to have activity spikes lasting about 2 minutes, and I think this might be contributing to slow printing for users and issues with deploying drivers on shared devices. We mainly use Konica Minolta MFPs and have Papercut set up. The spikes appear to be linked to the Print Spooler, creating several temporary files (KCM****.tmp). It seems like Windows could be querying the printers, but I'm not sure how to confirm this.
So far, I've tried turning off Print Isolation for all drivers and disabling SNMP, as well as reinstalling the same drivers, though I'm not sure if that made a difference. I haven't gone back to older drivers yet due to the hassle, given that we have around 40 MFPs with various settings. I'm hoping to hear if anyone else has faced something similar and if there's a way to alleviate this issue—maybe reduce the frequency of the checks to once an hour or once a day?
5 Answers
We switched our print services to PrinterLogic a few years back, and the number of printer-related issues dropped significantly. We used to receive daily tickets for all kinds of problems, but since the switch, we've only had about 5 tickets in 3 years. It’s been a game-changer for us!
Have you checked if an antivirus program or Windows Defender might be running scans during those spikes? That could be causing the performance hit.
It might not be DNS related as you might think. We use Papercut along with HP universal and Ricoh universal drivers, and it works great for us without any issues.
Two things might be causing the spikes. If users have the 'print immediately' option selected, their PCs will try to connect directly to the printer and bypass the print server, which can cause issues, especially in double firewall setups. Also, due to the print nightmare exploit, drivers might not deploy without admin rights, which complicates things if you're using vendor-specific drivers. We've dealt with this by switching to universal drivers on the print server whenever possible and deploying them across all machines.
I’ve seen people complain about slow printing—it’s noticeably slower than printing directly to the printer with the same drivers. We’ve also had slow logins, particularly when logins coincide with those spikes. Here’s a CPU activity graph I've been tracking that shows the fluctuations.
+1 for PrinterLogic! Definitely worth considering.