[PATCH] python indent bugfix in dns_hub.py

Tim Beale timbeale at catalyst.net.nz
Tue Jan 22 04:37:26 UTC 2019

I'm not sure this patch really helps. My CI builds are still hitting the
error, it's just now it looks more like this:

> INFO 2019-01-22 17:12:44,430 pid:95958 bin/python/samba/join.py #103:
> Finding a writeable DC for domain 'SAMBA2000.EXAMPLE.COM'
> dns_hub: sending to address ('', 55960) for name
> _ldap._tcp.SAMBA2000.EXAMPLE.COM
> Error sending to address ('', 55960) for name
> _ldap._tcp.SAMBA2000.EXAMPLE.COM: [Errno 113] No route to host
> ERROR: Failed to find a writeable DC for domain
> 'SAMBA2000.EXAMPLE.COM': The object name is not found.
> Join failed

E.g. https://gitlab.com/catalyst-samba/samba/-/jobs/148978423

I think the join code is essentially calling finddcs_cldap() using the
domain name, which does a DNS lookup. Because the DNS lookup fails, so
does the join, and so the testenv can't be created.

If I put a 2 second sleep in dns_transaction_udp(), then I see the exact
same failure locally. Which suggest to me we're hitting this because
finddcs_cldap() has timed out, samba-tool has given up and exited, and
so the client no longer exists for sock.sendto() to send the DNS
response back to.

I'm guessing this could either be happening because the DC's DNS server
process is slow to process the query (because the DC only just started
up), or the CI machine is just heavily loaded and so it takes a while to
proxy the DNS query/response back and forth between the DC and client.

On 21/01/19 7:06 AM, Andrew Bartlett wrote:
> On Fri, 2019-01-18 at 12:49 -0800, Jeremy Allison via samba-technical
> wrote:
>> On Sat, Jan 19, 2019 at 09:44:59AM +1300, Douglas Bagnall wrote:
>>> What we have managed so far is to crash with the correct message
>>> when
>>> a client disappears, but a more conventional approach might be to
>>> note
>>> the problem and keep going. This patch wants to do that.
>> LGTM. I was the one who added the raise in the first place,
>> and I'm OK with removing it. Although I'd really like to
>> know *why* the client is disappearing in the tests (which
>> is why I added the raise initially to try and track it down).
>>> I am not sure what the underlying cause is. The quoted example
>>> from https://gitlab.com/samba-team/devel/samba/-/jobs/147653485
>> It's intermittent, that's the problem. Mostly it doesn't
>> happen, I'd really like to know why it sometimes does :-(.
> Me too.  But I've pushed the patch anyway.
> Andrew Bartlett

More information about the samba-technical mailing list