Re: Peer authentication failure with postgres tests on ppc64 only?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 28/07/2019 21:08, Tom Hughes wrote:
On 28/07/2019 20:00, Kevin Fenzi wrote:
On 7/28/19 1:52 AM, Tom Hughes wrote:
I have a number of packages which use %postgresql_tests_run to start
a postgres server in %check and they all seem to have failed during
the mass rebuild, for example:

   http://koji.fedoraproject.org/koji/buildinfo?buildID=1331249
   http://koji.fedoraproject.org/koji/buildinfo?buildID=1331278

The failure is in %postgresql_tests_run when, after it has started
the server it tries to create a database:

+ createdb -h /tmp mockbuild --owner mockbuild -p 36884
BUILDSTDERR: createdb: could not connect to database template1: FATAL:
Peer authentication failed for user "mockbuild"

That basically means that createdb connected to the server using a
unix domain socket but the server didn't see the credentials on the
socket identifying the remote party as mockbuild correctly.

This is only happening on ppc64le so I'm guessing at some sort of
kernel problem on the ppc64le builders unless somebody has another
suggestion as to what is going on?

Huh, not sure what it could be.

Does it also happen on the ppc64le test instance?

It does, yes. I just ran a test build under mock on the test instance
and it failed in the same way.

I've reproduced this manually in mock on the test machine, and
straced the server and compared it to x86_64 and there is a key
difference. On ppc64le the connection packet is read with recv:

4761 recv(10, "\0\0\0D\0\3\0\0user\0mockbuild\0database\0"..., 8192, 0) = 68

while on x86_64 it is read with recvfrom:

24657 recvfrom(9, "\0\0\0>\0\3\0\0user\0thh\0database\0postgr"..., 8192, 0, NULL, NULL) = 62

and it then goes on to fetch the peer credentials:

24657 getsockopt(9, SOL_SOCKET, SO_PEERCRED, {pid=24656, uid=2067, gid=103}, [12]) = 0

while ppc64le fetches the socket low water mark:

4761  getsockopt(10, SOL_SOCKET, SO_SNDLOWAT, [1], [12->4]) = 0

after which it errors.

So it looks like some sort of bug with how postgres has been
built on ppc64le in fact.

Tom

--
Tom Hughes (tom@xxxxxxxxxx)
http://compton.nu/
_______________________________________________
devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx
To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx
Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/devel@xxxxxxxxxxxxxxxxxxxxxxx




[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Index of Archives]     [Fedora Announce]     [Fedora Users]     [Fedora Kernel]     [Fedora Testing]     [Fedora Formulas]     [Fedora PHP Devel]     [Kernel Development]     [Fedora Legacy]     [Fedora Maintainers]     [Fedora Desktop]     [PAM]     [Red Hat Development]     [Gimp]     [Yosemite News]

  Powered by Linux