Failure to create ipv6 socket

Server operators
1

My NTP server on a VPS dies occasionally (twice now) … which results in a service interruption. See pool.ntp.org: Statistics for 203.57.51.100

Looking in the syslog shows:

Aug 21 03:29:56 combine-window dhclient[507]: DHCPREQUEST for 203.57.51.100 on eth0 to 103.16.128.248 port 67 (xid=0x4311e9a7)
Aug 21 03:29:56 combine-window dhclient[507]: DHCPACK of 203.57.51.100 from 103.16.128.248 (xid=0xa7e91143)
Aug 21 03:29:56 combine-window dhclient[507]: bound to 203.57.51.100 – renewal in 274 seconds.
Aug 21 03:30:01 combine-window CRON[23839]: (nobody) CMD (/usr/local/bin/mpanel-memory-graph)
Aug 21 03:31:20 combine-window ntpd[814]: bind(25) AF_INET6 2404:9400:1:0:216:3eff:fef2:1a8a#123 flags 0x11 failed: Cannot assign requested address
Aug 21 03:31:20 combine-window ntpd[814]: unable to create socket on eth0 (6) for 2404:9400:1:0:216:3eff:fef2:1a8a#123
Aug 21 03:31:20 combine-window ntpd[814]: failed to init interface for address 2404:9400:1:0:216:3eff:fef2:1a8a
Aug 21 03:31:22 combine-window ntpd[814]: Listen normally on 7 eth0 [2404:9400:1:0:216:3eff:fef2:1a8a]:123
Aug 21 03:31:22 combine-window ntpd[814]: new interface(s) found: waking up resolver
Aug 21 03:31:29 combine-window ntpd[814]: Deleting interface #7 eth0, 2404:9400:1:0:216:3eff:fef2:1a8a#123, interface stats: received=0, sent=0, dropped=0, active_time=7 secs

I currently prevent the problem by making NTP ipv4 only.

Can anyone suggest what causes this problem???

2

Hi @richard, looks like a problem assigning an IP address via DHCP:

Googling a bit it seems to be a known issue for a while! (e.g. Bug #700492 “ntp complains about ipv6 errors every 5 minutes” : Bugs : ntp package : Ubuntu)

Does the server have IPv6 available? If not then one suggested workaround is to disable IPv6 - this does it for the entire machine (ntpd process bind() cannot assign requested address - Red Hat Customer Portal)

3

Elljay,

Thanks for the links… I have seen most of them … particularly the ‘every 5 minutes’ discussion. My problem is not like that … the server runs for days and then hiccups.

I think in both cases it was after a DHCPD update and one theory is that NTPD catches DHCP in mid renewal …

Tomorrow I will try testing the ipv6 connectivity of the VPS.

Will report back the results.

Thanks, Richard.

4

The virtual server did not have ipv6 turned on … so there was only a link local ipv6 address. My error for not knowing this.

It is now on and I can ping6 across the country. Will enable ipv6 on NTP and see if the problem recurs…

Richard.

1 Like
5

Don’t you just hate that :grinning: Hopefully it will be ok now :+1:

au looks like ti could do with some more IPv6 servers :slightly_smiling_face: pool.ntp.org: NTP Servers in Australia, au.pool.ntp.org

1 Like