Bug 11054

Shouldn't we be building our own gpg-native to perform any package signing, rather than relying on host utils?

(In reply to comment #1)
> Shouldn't we be building our own gpg-native to perform any package signing,
> rather than relying on host utils?

Yes we are but seems a problem of configuration at ~/.gpg, i suggest the test to create a isolated configuration.

The package signing classes have a GPG_PATH variable:

"Optional variable for specifying the gnupg "home" directory"

perhaps the tests could make use of that?

Can you try if the issue still happens with this branch? 
https://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/log/?h=akanavin/dnf-rpm4

Markus Lehtonen and I did some fixing specifically to make gpg work again, including Ubuntu. But it's done on top of rpm4/dnf, which is coming soon anyway.

(In reply to comment #3)
> The package signing classes have a GPG_PATH variable:
> 
> "Optional variable for specifying the gnupg "home" directory"
> 
> perhaps the tests could make use of that?

The tests are already setting up their private gnupg home directory (see the --homedir parameter above), this is not an issue.

Yes, this will be fixed with Ale(In reply to comment #4)
> Can you try if the issue still happens with this branch? 
> https://git.yoctoproject.org/cgit/cgit.cgi/poky-contrib/log/?h=akanavin/dnf-
> rpm4
> 
> Markus Lehtonen and I did some fixing specifically to make gpg work again,
> including Ubuntu. But it's done on top of rpm4/dnf, which is coming soon
> anyway.

Yes, this will be fixed with Alexander's dfn/rpm4 patchset. I've tested it specifically on Ubuntu 16.10 and it works.

Tested on GDC local AB were the failure was found and the 'gpg' failure is not present anymore.

This is still broken in master. Looks like the dnf patchset will not be merged until 2.4 so I'll fix this separately.

I'd say we need to wait for an official decision by RP; I've sent the v2 of the dnf patch with everything fixed, so it may yet find its way to master.

Uh, I already sent a patchset for review. But, now we have a fix in any case. Let RP decide :)

https://patchwork.openembedded.org/series/5471/

I don't mind that, if your fixes are merged, I can drop a few patches from the dnf patchset, which are only tangential to its purpose.

Fixed in master:
http://git.openembedded.org/openembedded-core/commit/?id=44a44b7e582a5a654baf21829d168568481c13d9

Selftest is failing on Ubuntu 17.04
2.2.2.rc2
All other workers are working correctly

======================================================================
ERROR [0.000s]: setUpClass (oeqa.selftest.signing.Signing)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/ab/autobuilder/yocto-worker/nightly-oe-selftest-ubuntu1704/build/meta/lib/oeqa/selftest/signing.py", line 30, in setUpClass
    runCmd('gpg --homedir %s --import %s %s' % (cls.gpg_dir, cls.pub_key_path, cls.secret_key_path))
  File "/home/ab/autobuilder/yocto-worker/nightly-oe-selftest-ubuntu1704/build/meta/lib/oeqa/utils/commands.py", line 114, in runCmd
    raise AssertionError("Command '%s' returned non-zero exit status %d:\n%s" % (command, result.status, result.output))
AssertionError: Command 'gpg --homedir /tmp/oeqa-signing-hc7m5doc --import /home/ab/autobuilder/yocto-worker/nightly-oe-selftest-ubuntu1704/build/meta-selftest/files/signing/key.pub /home/ab/autobuilder/yocto-worker/nightly-oe-selftest-ubuntu1704/build/meta-selftest/files/signing/key.secret' returned non-zero exit status 2:
gpg: keybox '/tmp/oeqa-signing-hc7m5doc/pubring.kbx' created
gpg: /tmp/oeqa-signing-hc7m5doc/trustdb.gpg: trustdb created
gpg: key 7B31316B5D64AD52: public key "testuser (nocomment) <testuser@testemail.com>" imported
gpg: key 7B31316B5D64AD52: "testuser (nocomment) <testuser@testemail.com>" not changed
gpg: key 7B31316B5D64AD52/7B31316B5D64AD52: error sending to agent: Inappropriate ioctl for device
gpg: error building skey array: Inappropriate ioctl for device
gpg: Total number processed: 2
gpg:               imported: 1
gpg:              unchanged: 1
gpg:       secret keys read: 1

----------------------------------------------------------------------

Please open a new bug, as this is a different issue.