Netgate Discussion Forum
    • Categories
    • Recent
    • Tags
    • Popular
    • Users
    • Search
    • Register
    • Login

    A1SAi-2750F nmbclusters limit reached

    Scheduled Pinned Locked Moved Hardware
    11 Posts 6 Posters 3.6k Views
    Loading More Posts
    • Oldest to Newest
    • Newest to Oldest
    • Most Votes
    Reply
    • Reply as topic
    Log in to reply
    This topic has been deleted. Only users with topic management privileges can see it.
    • O
      ocpistol
      last edited by

      Just Built a second router as a backup and I put it as my main to do some stuff on my other one and I am getting kern.ipc.nmbclusters limit reached.

      It wont let me log into the web config
      c2750.png
      c2750.png_thumb

      1 Reply Last reply Reply Quote 0
      • Q
        q54e3w
        last edited by

        Open shell, edit boot/loader.conf config file to increase MBUFs (say 512000) and restart

        1 Reply Last reply Reply Quote 0
        • O
          ocpistol
          last edited by

          I dont see that in the file.

          pfsense-loader.png
          pfsense-loader.png_thumb

          1 Reply Last reply Reply Quote 0
          • Q
            q54e3w
            last edited by

            Just add the line….I'm on a tablet so cant check exact spelling etc but it should be something like

            kern.ipc.nmbclusters="512000"

            Save and restart.....you should be good then.

            1 Reply Last reply Reply Quote 0
            • O
              ocpistol
              last edited by

              Seems to be working thank yoU!

              1 Reply Last reply Reply Quote 0
              • H
                heper
                last edited by

                don't add it to /boot/loader.conf …. it'll get overwritten by updates.

                https://doc.pfsense.org/index.php/Tuning_and_Troubleshooting_Network_Cards

                be sure to add it to /boot/loader.conf.local

                1 Reply Last reply Reply Quote 0
                • O
                  ocpistol
                  last edited by

                  I did this, Thank you very much for the help!

                  1 Reply Last reply Reply Quote 0
                  • H
                    Harvy66
                    last edited by

                    What causes more of these to get used? I have never seen my usage budge since I first installed PFSense. I changed it to 1,048,576 because a PFSense doc recommended this for high memory systems with 1Gb+ interfaces, and that increased my usage by a small amount.

                    Then I hear about people running out of them. Is it because they aren't being returns quickly enough?

                    1 Reply Last reply Reply Quote 0
                    • Q
                      q54e3w
                      last edited by

                      high CPU core count for one thing….

                      1 Reply Last reply Reply Quote 0
                      • J
                        jasonlitka
                        last edited by

                        @Harvy66:

                        What causes more of these to get used?

                        The Intel igb driver, by default, sets up 1 queue per CPU core per NIC.  The C2750/2758 Rangeley boxes have 8 cores & 4 NICs.

                        I ran into the same issue with my boxes at work which have (8) logical cores, (12) igb NICs, and (2) ix NICs.  You can either restrict the number of cores, the number of queues, or raise nmbclusters.

                        I can break anything.

                        1 Reply Last reply Reply Quote 0
                        • ?
                          Guest
                          last edited by

                          @Jason:

                          @Harvy66:

                          What causes more of these to get used?

                          @Jason:

                          The Intel igb driver, by default, sets up 1 queue per CPU core per NIC.

                          … in pfSense 2.2.    (Which is the correct default.)

                          @Jason:

                          The C2750/2758 Rangeley boxes have 8 cores & 4 NICs.

                          C2750 is Avoton, not Rangeley.

                          @Jason:

                          I ran into the same issue with my boxes at work which have (8) logical cores, (12) igb NICs, and (2) ix NICs.  You can either restrict the number of cores, the number of queues, or raise nmbclusters.

                          raising nmbclusters is, by far, the best approach here.

                          1 Reply Last reply Reply Quote 0
                          • First post
                            Last post
                          Copyright 2025 Rubicon Communications LLC (Netgate). All rights reserved.