DNS issues in selftest
garming at catalyst.net.nz
Sun Jun 19 21:16:51 UTC 2016
I haven't been able to reproduce this on its own on sn-devel or on my
machine. This test itself seems to indicate that the DNS server gave a
response, but the forwarder which was set did not. As far as I can tell,
there's only two ways this can happen:
1. The DNS forwarder did not bind to the socket correctly, possibly
either in the time span allocated (30 seconds) or some other permission
2. The DNS server did not get the response from the forwarder, despite
the forwarder being up. Either the message took too long in transit, was
dropped, or the DNS forwarder itself took too long.
All these situations seem rather unlikely. Unless, there's some other
case I'm not considering.
One oddity I also noticed about this test is that that the logs seem
rather consistent with logs where the forwarder tests have already just
finished. And this is also the first test of the forwarder tests. It
would be incredibly strange to find that this was somehow running twice,
unintendedly (although that might explain the forwarder unable to bind
with a possible tearDown failure). I'll change a few lines in the test
to see if I can get some more information, but it all seems very strange
to me and an issue with the environment (not the actual forwarder).
On 19/06/16 00:30, Andreas Schneider wrote:
> Hi Garming,
> as you worked on DNS lately, I think we want to see more information if the
> test fails here to find out why it is out of range.
> [1578(9717)/1974 at 2h37s] samba.tests.dns_forwarder(fl2003dc:local)
> REASON: Exception: Exception: Traceback (most recent call last):
> line 465, in test_cname
> self.assertEqual('forwarder1', data.answers.rdata)
> IndexError: list index out of range
> This is from an autobuild mail.
> -- andreas
More information about the samba-technical