Suggest you test with 2.91 RC2. From: Dnsmasq-discuss <dnsmasq-discuss-boun...@lists.thekelleys.org.uk> On Behalf Of Bryant Eadon Sent: Sunday, February 23, 2025 2:05 AM To: dnsmasq-discuss@lists.thekelleys.org.uk Subject: [Dnsmasq-discuss] Tracing a dnsmasq bug - hot loop - possibly DHCP6 lease management
Hello, I am tracing an issue where dnsmasq 2.86 appears to get stuck in a loop using all compute resources for the core it's on, and becomes unresponsive to DHCP4, DHCPv6 and DNS requests. I am working on replicating the problem, and want to check to see if anyone on this list has seen the behavior before -- -- Details -- On the local LAN I am seeing 3 devices in DHCPv6 that are requesting IPV6 addresses every 10 minutes, even though their expiry time is advertised as 8 hours. I will have some difficulty finding the exact devices. It appears the DHCPv6 lease file continues tracking all of the leases on this openwrt based router. A subset of these tracked requests are here, from the lease file - 1740156067 1484488511 2600:4040:a92e:2700::ffcd * 00:03:00:01:f0:09:0d:cb:26:08 1740156067 1484488511 fd30:fc9a:a298::f8a3 * 00:03:00:01:f0:09:0d:cb:26:08 1740156067 1375367042 2600:4040:a92e:2700::ffcc * 00:03:00:01:f0:09:0d:cb:27:52 1740156067 1375367042 fd30:fc9a:a298::f8a2 * 00:03:00:01:f0:09:0d:cb:27:52 1740155562 1744997917 2600:4040:a92e:2700::ffcb * 00:03:00:01:f0:09:0d:cb:26:08 1740155562 1744997917 fd30:fc9a:a298::f8a1 * 00:03:00:01:f0:09:0d:cb:26:08 1740155561 729442040 2600:4040:a92e:2700::ffca * 00:03:00:01:f0:09:0d:cb:27:52 1740155561 729442040 fd30:fc9a:a298::f8a0 * 00:03:00:01:f0:09:0d:cb:27:52 1740155561 1831382606 2600:4040:a92e:2700::ffc9 * 00:03:00:01:f0:09:0d:cb:25:b0 1740155561 1831382606 fd30:fc9a:a298::f89f * 00:03:00:01:f0:09:0d:cb:25:b0 1740155018 995327705 2600:4040:a92e:2700::ffc8 * 00:03:00:01:f0:09:0d:cb:25:b0 1740155018 995327705 fd30:fc9a:a298::f89e * 00:03:00:01:f0:09:0d:cb:25:b0 So, this is 48 entries (identical) in the dhcpv6 lease file per host that does this. There are 2 entries, per host (96 entries) .. and there are 3 hosts on the LAN doing this -- bringing the entries to 288 tracked leases, excluding the IPV4 leases too. My dnsmasq config file is attached ( dnsmasq-configfile.txt ). The only parameter I imagine is related : dhcp-range=set:lan,192.168.2.40,192.168.2.254,255.255.255.0,8h Notably - this is not a DHCPv6 setting, but it does show the 8h lease time I am seeing in the DHCPv6 leases , AND the parameters here have a limit of 215 addresses these hosts are pushing (the same) number of leases over 288 within an 8hour span. Has anyone seen this before ? Why are the IPv6 leases not replaced in the lease file when the same DUID requests addresses well before the end of the IPv6 lease time ? *Bonus* -- I've attached a pcap file for the DHCPv6 requests I think could be central to this problem. These packets arrived much later of course, well after I've experienced this problem, but may give us some insight to what's happening. Thank you, Bryant Eadon ᐧ
_______________________________________________ Dnsmasq-discuss mailing list Dnsmasq-discuss@lists.thekelleys.org.uk https://lists.thekelleys.org.uk/cgi-bin/mailman/listinfo/dnsmasq-discuss
