OPNsense 24.7: "ping: sendto: No buffer space available" with Tor relays behind OPNsense

[neelchauhan]

I'm running OPNsense 24.7 on a Minisforum MS-01 and my connection runs Tor relays. The Tor relays are on two HPE ML110 Gen11 servers. I average at ~150 Mbps throughput and maintain a rather large state table.

After a few hours of uptime, I get this error:

ping: sendto: No buffer space available

And in turn, the WAN is completely down until I reboot. If I use the integrated I226 NICs, it happens after an hour. Adding a Broadcom BCM5720 NIC delays it to about 12 hours but it still eventually crashes. This also happened on a Deciso DEC2752 but only within a few minutes.

I'm using Frontier FiberOptic in Stamford, Connecticut, but since Connecticut is ex-AT&T territory we have 802.1X as we have AT&T Fiber with Frontier billing/support and no IPv6. MikroTik never had any issues with bypassing, but noise was an issue with the CCR2004 series. I could switch to XGS-PON since Frontier gave up on 802.1X here but didn't bother with a needed tech visit. Optimum is only DOCSIS here and GoNetSpeed is unavailable.

Note: I'm going to be moving to NYC in a week, and will have Verizon Fios without the 802.1X bullshit so I won't need this forever. My family is keeping the Stamford home and I'll put a smaller MikroTik box here or even put the Frontier Arris back.

[owenthewizard]

This doesn't happen without bypass? What method are you using, in detail?

[neelchauhan]

I was using the opnatt non-supplicant bypass. I was able to fix this by disabling logging, since then I was up 24+ hours.

[neelchauhan]

Reopening since it still re-appeared but took much longer