MSSQL Response Time for SQL DB hosts due to Npcap 1.31

[chaanth]

Describe the bug

• We are querying where time_taken (response time) > 10 seconds for SQL DB hosts with Npcap 1.31.
• You can see a big spike every hour of 100+ events which goes up depending on how busy our system is.
• You can't tell from here but the time increments by 30 seconds each hour.
• The other indicator is when we get cluster disconnect errors in the SQL Server event log. We have 5 servers in a cluster hosting the database, and this error indicates they timed out on their heartbeat connection. These don’t happen every hour, just several times per day but they happen at the times when the slowness happens.
• Upgrading from the Winpcap driver to an Npcap driver has still not resolved our issue
• Initially, the spikes correlated with metric reporting time so we changed the interval to test and nothing changed.
• After removing all core components without success, we removed the pcap driver entirely, and the issue went away.

Diagnostic information Windows Server 2016 Standard 1607, ver. 10.0.14393

Additional context Attached DiagReport, Install.log, and NPFInstall log

[chaanth]

Can this be looked into?

[guyharris]

Upgrading from the Winpcap driver to an Npcap driver has still not resolved our issue

I.e., this is an issue with both the WInpcap driver and the Npcap 1.31 driver.

I think a lot of work has been done on the driver since 1.31; what happens with newer versions of Npcap?

[ChrisCzajka]

the issue is still being observed with npcap 1.71 just as before