Python 2.7 root buildbot showing errors
The first build my new root buildbot did showed errors in the 2.7 test suite, but I thought little of it as quite a few other 2.7 buildbots are showing red, too. But it seems they're showing different errors, so there might be something wrong with the setup. http://buildbot.python.org/all/builders/AMD64%20Debian%20root%202.7/builds/3... First off, it's complaining about being unable to build _curses (lacking curses.h). Is that a mandatory prereq that I should install, or should Python be compatible with not having it? Then further down, several SSL tests attempt: s.connect_ex(("svn.python.org", 444))) and get back EAGAIN when they're expecting ECONNREFUSED. Possibly my firewall's delaying things somewhat and it's timing out with a signal; when I try manually, the connection times out. Are these failures a problem? Should they be fixed? The 3.x builds are all coming up green. ChrisA
On 06.01.2014 08:09, Chris Angelico wrote:
Then further down, several SSL tests attempt:
s.connect_ex(("svn.python.org", 444)))
and get back EAGAIN when they're expecting ECONNREFUSED. Possibly my firewall's delaying things somewhat and it's timing out with a signal; when I try manually, the connection times out.
Are you running the VM on Windows? I've seen similar issues on Windows and Windows as host platform for VMs: http://bugs.python.org/issue19919 Christian
On Mon, Jan 6, 2014 at 7:11 PM, Christian Heimes <christian@python.org> wrote:
On 06.01.2014 08:09, Chris Angelico wrote:
Then further down, several SSL tests attempt:
s.connect_ex(("svn.python.org", 444)))
and get back EAGAIN when they're expecting ECONNREFUSED. Possibly my firewall's delaying things somewhat and it's timing out with a signal; when I try manually, the connection times out.
Are you running the VM on Windows? I've seen similar issues on Windows and Windows as host platform for VMs:
No, it's Debian Wheezy inside Debian Wheezy; though the outer system is a somewhat messy one (I installed Wheezy before it was stable, and compiled my own ALSA drivers and a few other things). But it could well be that same issue, as it seems to involve NAT. What's the policy on backporting patches to tests onto 2.7? ChrisA
On 06.01.2014 09:22, Chris Angelico wrote:
No, it's Debian Wheezy inside Debian Wheezy; though the outer system is a somewhat messy one (I installed Wheezy before it was stable, and compiled my own ALSA drivers and a few other things).
But it could well be that same issue, as it seems to involve NAT. What's the policy on backporting patches to tests onto 2.7?
Interesting, maybe it's a general NAT issue? So far I have seen the issue on Windows only. What kind of VM are you using? I'm using virtualbox for my Windows VMs. Just backport the test fixes. Test fixes and new tests are not new feature so you are always allowed to add new tests or fix existing tests. Christian
On Mon, Jan 6, 2014 at 7:58 PM, Christian Heimes <christian@python.org> wrote:
Interesting, maybe it's a general NAT issue? So far I have seen the issue on Windows only. What kind of VM are you using? I'm using virtualbox for my Windows VMs.
It's Oracle VirtualBox v4.2.20 r90963.
Just backport the test fixes. Test fixes and new tests are not new feature so you are always allowed to add new tests or fix existing tests.
Okay. I don't have write access, myself. ChrisA
On Mon, Jan 6, 2014 at 2:09 AM, Chris Angelico <rosuav@gmail.com> wrote:
The first build my new root buildbot did showed errors in the 2.7 test suite, but I thought little of it as quite a few other 2.7 buildbots are showing red, too. But it seems they're showing different errors, so there might be something wrong with the setup.
http://buildbot.python.org/all/builders/AMD64%20Debian%20root%202.7/builds/3...
First off, it's complaining about being unable to build _curses (lacking curses.h). Is that a mandatory prereq that I should install, or should Python be compatible with not having it?
Yes, curses should be considered entirely optional so the tests should still pass (as long as the build doesn't error out then the compiler message about not being able to build curses is not critical). -Brett
Then further down, several SSL tests attempt:
s.connect_ex(("svn.python.org", 444)))
and get back EAGAIN when they're expecting ECONNREFUSED. Possibly my firewall's delaying things somewhat and it's timing out with a signal; when I try manually, the connection times out.
Are these failures a problem? Should they be fixed? The 3.x builds are all coming up green.
ChrisA _______________________________________________ Python-Dev mailing list Python-Dev@python.org https://mail.python.org/mailman/listinfo/python-dev Unsubscribe: https://mail.python.org/mailman/options/python-dev/brett%40python.org
participants (3)
-
Brett Cannon -
Chris Angelico -
Christian Heimes