[Bug 1771353] Re: Server feature level is now lower than when we began our transaction. Restarting with new ID

[Dimitri John Ledkov]

On 3 August 2018 at 14:22, Steve Dodd <1771353 at bugs.launchpad.net> wrote:
> @xnox:
>
> Yes, on further investigation it looks like these might well be
> different bugs - OTOH the repeated issuance of queries to the upstream
> server, and the confusion about the server capabilities is very similar,
> which suggests a problem in the same code path may be being triggered. I
> will create a new bug once I am sure they are different.
>

Yeah, a lot of those debug output from resolved is just that debug
output, and it looks very similar for even regular normal operations.
E.g. the dns server on router is crap and I have transactions
restarting, lowering the supported level, then later it attempts to up
the supported level again only go back down again... So restarting of
transactions and changing ids is "normal" as it's simply "verbose".

> Re. libnss-resolve, the man page still says:
>
> "it is strongly recommended that local programs use the glibc NSS or bus
> APIs instead [of the stub resolver]"
>

That sounds like it predates implementation of stub-resolv.conf
generator with search stanzas and other improvements to the local stub
listener for large query sizes.


> But I do understand there are potential issues - might be nice to have
> the Ubuntu "position" documented (top of resolv.conf, perhaps).
>

I'm working on a blog post series / wiki page, and a talk for FOSDEM
about it. But it has been going slow...


> Simon Kelley is looking at the dnsmasq issues.
>

cool!

-- 
Regards,

Dimitri.

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to systemd in Ubuntu.
https://bugs.launchpad.net/bugs/1771353

Title:
  Server feature level is now lower than when we began our transaction.
  Restarting with new ID

Status in systemd package in Ubuntu:
  Confirmed

Bug description:
  systemd 237-3ubuntu10 from bionic

  There is no /etc/netplan/* file.

  I'm getting some sort of a loop when systemd-resolved tries to resolve
  the hostname of my canonistack instance:

  ubuntu at bip:~$ systemd-resolve bip.lcy02.canonistack
  bip.lcy02.canonistack: resolve call failed: All attempts to contact name servers or networks failed

  systemd-resolved debug output loops over like this:
  Transaction 36341 is now 19121.
  Cache miss for bip.lcy02.canonistack IN A
  Transaction 19121 for <bip.lcy02.canonistack IN A> scope dns on */*.
  Using feature level UDP+EDNS0 for transaction 19121.
  Sending query packet with id 19121.
  Processing incoming packet on transaction 19121. (rcode=SUCCESS)
  Server feature level is now lower than when we began our transaction. Restarting with new ID.
  Transaction 19121 is now 17157.
  Cache miss for bip.lcy02.canonistack IN A
  Transaction 17157 for <bip.lcy02.canonistack IN A> scope dns on */*.
  Using feature level UDP+EDNS0 for transaction 17157.
  Sending query packet with id 17157.
  Processing incoming packet on transaction 17157. (rcode=SUCCESS)
  Server feature level is now lower than when we began our transaction. Restarting with new ID.
  ...

  Packet capture shows a loop like this:
  13:38:32.293524 IP 10.55.32.54.53786 > 10.55.32.1.53: 46096+ [1au] A? bip.lcy02.canonistack. (50)
  13:38:32.293753 IP 10.55.32.1.53 > 10.55.32.54.53786: 46096* 1/0/0 A 10.55.32.54 (55)
  13:38:32.294026 IP 10.55.32.54.53786 > 10.55.32.1.53: 64494+ [1au] A? bip.lcy02.canonistack. (50)
  13:38:32.294249 IP 10.55.32.1.53 > 10.55.32.54.53786: 64494* 1/0/0 A 10.55.32.54 (55)

  I'm going to attach the full debug output, and the pcap file.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/systemd/+bug/1771353/+subscriptions