Status Queues Drops keep resetting on 2.4.5-Release



  • Watching Status->Queues when network somewhat busy, I can watch the drops slowly going up for a while, then suddenly the bandwidth column shows all negative numbers (for one 1-sec refresh) then the drops have all been zeroed out. I never noticed anything like this before 2.4.5-Release.

    Has anyone else seen this? Not sure what info would be relevant. 10 interfaces having up to 10 queues. All PRIQ. Upload speed (10Mbit) was pegged at the time over an OpenVPN P2P interface. 2.4.5 was an upgrade from 2.4.4-P2 2 days ago. Hardware on both ends is APU2C4. I was watching a Synology Hyper Backup. WAN is alone on a port, not on a VLAN. Traffic from Synology box enters a different port over a VLAN. Everything seems to be working fine, just Traffic Shaper Queue stats resetting.

    Thanks!

    -Steve


  • Rebel Alliance Developer Netgate

    Do you see anything in the system log around the time it resets? Or the gateways log?



  • @jimp Hmm, yes, DNS Resolver log shows filterdns doing merge_config: configuration reload followed by Adding a bunch of hosts. Every 15 minutes. Coincides exactly with Queue status reset.

    I don't see anything else happening in the logs at that time.


  • Rebel Alliance Developer Netgate

    The queue stats are likely being reset during the filter reload. The updates are likely from something like pfBlockerNG updating DNS-based lists and are probably not something you can avoid without reducing the effectiveness of whichever feature is taking that action.



  • @jimp Thanks Jim. The only package installed is Openvpn client exporter. Anything I can do to narrow down what would be causing the filter reloads? They are happening at the 0, 15, 30 and 45 minute marks.



  • Jim, I installed cron and see this: /etc/rc.filter_configure_sync scheduled every 15 minutes, which you suggest in another thread is probably because of using a schedule.

    So I guess if you use a schedule, Queue stats are cleared every 15 minutes.

    Thanks,

    -Steve


Log in to reply