Bug #267
DNS doesn't appear to work on some interfaces
| Status: | Closed | Start date: | 09/10/2011 | |
|---|---|---|---|---|
| Priority: | Urgent | Due date: | ||
| Assignee: | % Done: | 0% |
||
| Category: | - | Spent time: | - | |
| Target version: | 1st Public Cerowrt release |
Description
The report:
This is on an 802.11g (sw00) channel, since there seems to be an issue with DNS on the .11a (sw10) channel. Interestingly enough, when I associate with the .11a guest (sw11) channel, the DNS works. (Go figure; would you like me to file a bug report on that?)
I've seen this too. bind9 is supposed to scan the available interfaces every two minutes, and it seems possible to trigger a new scan when a new interface comes up OR have some linux specific monitoring thing (inotify, kernel event?)
Certainly all the wireless interfaces are not active when named starts...
I also note that this report is AFTER we upgraded to bind 9.8.1....
Related issues
| related to Cerowrt - Bug #262: poor wireless performance in smoketest rc6 | In Progress | 09/04/2011 |
History
Updated by Frank Horowitz over 1 year ago
OK, A closer look reveals that the interface sw10 does not have an IP address, even though gw10 (and all the others) do.
Possibly related are the following "bad number" error messages from logread:
Sep 14 08:59:53 Lirpo5 daemon.warn radvd4403: resetting ipv6-allrouters membership on sw00
Sep 14 08:59:54 Lirpo5 daemon.notice ntpd4468: ntpd 4.2.6p3@1.2290-o Fri Sep 9 22:12:28 UTC 2011 (1)
Sep 14 08:59:54 Lirpo5 daemon.notice ntpd4469: proto: precision = 6.288 usec
Sep 14 08:59:54 Lirpo5 daemon.debug ntpd4469: ntp_io: estimated max descriptors: 1024, initial socket boundary: 16
Sep 14 08:59:54 Lirpo5 daemon.err ntpd4469: unable to bind to wildcard address 0.0.0.0 - another process may be running - EXITING
Sep 14 08:59:54 Lirpo5 user.info sysinit: sh: bad number
Sep 14 08:59:54 Lirpo5 user.info sysinit: sh: bad number
Sep 14 08:59:54 Lirpo5 user.info sysinit: sh: bad number
Sep 14 08:59:54 Lirpo5 user.info sysinit: sh: bad number
Sep 14 08:59:54 Lirpo5 user.info sysinit: setting up led WAN LED (green)
Sep 14 08:59:54 Lirpo5 kern.debug kernel: ar71xx-wdt: enabling watchdog timer
Sep 14 08:59:54 Lirpo5 user.info sysinit: Polipo is starting...
Updated by Frank Horowitz over 1 year ago
Found it. The config stanza for sw10 in /etc/config/network had some funky bridging and ifname lines. Commenting them out brings back the IP number assignment, (and presumably DNS et al.)...
Updated by Dave Täht over 1 year ago
thx! I'll fix this in rc6-smoketest8
Updated by Dave Täht over 1 year ago
or rather, I would, if that was a file showing problems in it. But it doesn't. So maybe something about the autogeneration of wifi interfaces messes with this. I will test before releasing smoketest8....
Updated by Dave Täht over 1 year ago
- Status changed from New to Feedback
- Assignee set to Dave Täht
the syntax of the /etc/config/network file had grown much pickier about quoting than previous. The quotes are now liberally throughout and appear to work. Smoketest9 will contain the fixes.
Thank you for the spot! as this disabled dhcp entirely on the main wireless interfaces it was very bad...
Updated by Frank Horowitz over 1 year ago
No worries! "All bugs are shallow, etc. etc. etc." ;-)
Just confirming that rc6 has fixed the issue for me.
Updated by Dave Täht over 1 year ago
so I (or you) can close this bug?
Updated by Frank Horowitz over 1 year ago
- Status changed from Feedback to Closed