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

    BGP dies randomly, write error: Operation not permitted & No buffer space avail

    Scheduled Pinned Locked Moved Routing and Multi WAN
    2 Posts 1 Posters 515 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.
    • T
      TeknikL
      last edited by

      Errors in the log:
      Apr 21 05:43:29 bgpd[93927]: neighbor x.x.x.x (STORM): pfkey setup failed
      Apr 21 05:43:29 bgpd[93927]: writev (8/104): No buffer space available
      Apr 21 05:43:28 bgpd[93927]: writev (6/80): No buffer space available
      Apr 21 05:42:58 bgpd[93927]: neighbor x.x.x.x (STORM): state change Established -> Idle, reason: Fatal error
      Apr 21 05:42:58 bgpd[93927]: neighbor x.x.x.x (STORM): write error: Operation not permitted

      I'm running 2.2.4 and ipsec/bgpd on the same box (I was reading this is my issue when trying to update to 2.3.3)

      This happened after 9 hours from last reboot. it used to happen all the time and then stopped happening for a while then came back after the last reboot. I end up losing my BGP default gateway in the middle of the night which breaks all my subnets.

      Is there a way to restart the bgp service if it throws the error state change Established -> Idle as a stop-gap until I can determine the root cause?

      I looked for these settings which supposedly fixed the VPN/BGPD issues in 2.3.3 but I couldnt find them.
      net.inet.raw.maxdgram  131072   
      net.inet.raw.recvspace  1048576   
      net.raw.recvspace  1048576   
      net.raw.sendspace  1048576

      1 Reply Last reply Reply Quote 0
      • T
        TeknikL
        last edited by

        I think I figured this one out!

        My dang syslog was logging to a remote site over a tunnel, and the tunnel was flaky at random times so the syslogd was filling up the tcp buffer.

        I'll report back if it's stable for multiple days.

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