Hi Ed, > On 19 Jun 2018, at 17:24, Ed Kern (ejk) <e...@cisco.com> wrote: > > > >> On Jun 19, 2018, at 7:33 AM, Ole Troan <otr...@employees.org> wrote: >> >> Dave, et al, >> >> Yes, all these are indications of the verify job running a different (and >> cached) version of the Python VPP package than the one in build tree. > > yup it certainly is... > >> That stopped working as soon as I changed the in-tree one (and the caching >> system picking up and apparently being stuck with a broken version of it). >> The fix is of course to always use the in-tree one. > question for you…did you rev the version number for vpp_papi from 1.4 with > your changes? (because if you rev’d and it still stuck with the old thats a > larger problem)
I at least tried to rev it later, if not in first rev of patch. Reverting the patch also changed the version number of course. Which didn’t get it unstuck. > > I put that patch in as you have already seen for the python_deps because on > at least four different occasions we had builds going bad intermittently > because make test-sep pulls over sixteen (yes 16) different packages. > This was bad and it was breaking things. > So yes every three days I rebuild the cache (for make install-dep and make > test-dep) I build them into the base build image. > Any new requirements in-between runs of the cache (additions either of > packages or version changes) are still picked up with each and every run > (make install-dep and make test-dep are still run with each verify) Good! We could rev vpp_papi on every change, but at least in principle we might not want that. Since the vpp_papi is in the repo there is certainly an expectation that we can rev and develop that just like any other in-repo code. I think it’s perfectly fine that the dependencies are in the cache, as long as they get updated whenever there is change in requirement. But I don’t think vpp_papi should be. (At least not behave as if it was.) > >> So that local verify behavior matches jenkins one. > > Well this may not be true for dpdk (since those packages are not built every > time) it should be true for anything else. Id need to see your gerrit > to try and parse out why it verified and then failed post merge. Thanks > > > >> But we need someone with access to the verify build. While we’re at it, >> would be great to have that documented and accessible... >> > > separate conversation (but its one im more than happy to have) on what you > would have liked to have access to or documented to make this > problem faster or easier for you to track down and correct. > > > Im now going to poke into removing papi after test-dep so that we dont lose > the benefits of having the external deps cached but forcing the build of what > should > be built each and every time. In this case the vpp-papi. Yep. Thanks Cheers Ole > > Ed > > > > >> Cheers >> Ole >> >> On 19 Jun 2018, at 15:21, Dave Barach via Lists.Fd.Io >> <dbarach=cisco....@lists.fd.io> wrote: >> >>> See, for example, https://gerrit.fd.io/r/#/c/13061. >>> >>> This failure is almost certainly unrelated to the patch. We need to fix >>> this 90 seconds ago. >>> >>> Thanks... Dave >>> >>> >>> 13:02:17 >>> ============================================================================== >>> 13:02:17 ERROR: IP ACL test >>> 13:02:17 >>> ------------------------------------------------------------------------------ >>> 13:02:17 Traceback (most recent call last): >>> 13:02:17 File >>> "/w/workspace/vpp-verify-master-ubuntu1604/test/test_classifier.py", line >>> 310, in test_acl_ip >>> 13:02:17 self.create_classify_table('ip', >>> self.build_ip_mask(src_ip='ffffffff')) >>> 13:02:17 File >>> "/w/workspace/vpp-verify-master-ubuntu1604/test/test_classifier.py", line >>> 244, in create_classify_table >>> 13:02:17 current_data_offset=data_offset) >>> 13:02:17 File >>> "/w/workspace/vpp-verify-master-ubuntu1604/test/vpp_papi_provider.py", line >>> 2206, in classify_add_del_table >>> 13:02:17 'mask': mask}) >>> 13:02:17 File >>> "/w/workspace/vpp-verify-master-ubuntu1604/test/vpp_papi_provider.py", line >>> 160, in api >>> 13:02:17 reply = api_fn(**api_args) >>> 13:02:17 File "build/bdist.linux-x86_64/egg/vpp_papi/vpp_papi.py", line >>> 104, in __call__ >>> 13:02:17 return self._func(**kwargs) >>> 13:02:17 File "build/bdist.linux-x86_64/egg/vpp_papi/vpp_papi.py", line >>> 398, in f >>> 13:02:17 return self._call_vpp(i, msg, multipart, **kwargs) >>> 13:02:17 File "build/bdist.linux-x86_64/egg/vpp_papi/vpp_papi.py", line >>> 591, in _call_vpp >>> 13:02:17 b = msg.pack(kwargs) >>> 13:02:17 File "build/bdist.linux-x86_64/egg/vpp_papi/vpp_serializer.py", >>> line 320, in pack >>> 13:02:17 b += self.packers[i].pack(data[a], kwargs) >>> 13:02:17 File "build/bdist.linux-x86_64/egg/vpp_papi/vpp_serializer.py", >>> line 171, in pack >>> 13:02:17 b += self.packer.pack(e) >>> 13:02:17 File "build/bdist.linux-x86_64/egg/vpp_papi/vpp_serializer.py", >>> line 51, in pack >>> 13:02:17 return self.packer.pack(data) >>> 13:02:17 error: cannot convert argument to integer >>> 13:02:17 >>> 13:02:17 >>> ============================================================================== >>> 13:02:17 ERROR: Output IP ACL test >>> 13:02:17 >>> ------------------------------------------------------------------------------ >>> 13:02:17 Traceback (most recent call last): >>> 13:02:17 File >>> "/w/workspace/vpp-verify-master-ubuntu1604/test/test_classifier.py", line >>> 340, in test_acl_ip_out >>> 13:02:17 data_offset=0) >>> 13:02:17 File >>> "/w/workspace/vpp-verify-master-ubuntu1604/test/test_classifier.py", line >>> 244, in create_classify_table >>> 13:02:17 current_data_offset=data_offset) >>> 13:02:17 File >>> "/w/workspace/vpp-verify-master-ubuntu1604/test/vpp_papi_provider.py", line >>> 2206, in classify_add_del_table >>> 13:02:17 'mask': mask}) >>> 13:02:17 File >>> "/w/workspace/vpp-verify-master-ubuntu1604/test/vpp_papi_provider.py", line >>> 160, in api >>> 13:02:17 reply = api_fn(**api_args) >>> 13:02:17 File "build/bdist.linux-x86_64/egg/vpp_papi/vpp_papi.py", line >>> 104, in __call__ >>> 13:02:17 return self._func(**kwargs) >>> 13:02:17 File "build/bdist.linux-x86_64/egg/vpp_papi/vpp_papi.py", line >>> 398, in f >>> 13:02:17 return self._call_vpp(i, msg, multipart, **kwargs) >>> 13:02:17 File "build/bdist.linux-x86_64/egg/vpp_papi/vpp_papi.py", line >>> 591, in _call_vpp >>> 13:02:17 b = msg.pack(kwargs) >>> 13:02:17 File "build/bdist.linux-x86_64/egg/vpp_papi/vpp_serializer.py", >>> line 320, in pack >>> 13:02:17 b += self.packers[i].pack(data[a], kwargs) >>> 13:02:17 File "build/bdist.linux-x86_64/egg/vpp_papi/vpp_serializer.py", >>> line 171, in pack >>> 13:02:17 b += self.packer.pack(e) >>> 13:02:17 File "build/bdist.linux-x86_64/egg/vpp_papi/vpp_serializer.py", >>> line 51, in pack >>> 13:02:17 return self.packer.pack(data) >>> 13:02:17 error: cannot convert argument to integer >>> 13:02:17 >>> > -=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#9647): https://lists.fd.io/g/vpp-dev/message/9647 Mute This Topic: https://lists.fd.io/mt/22432813/21656 Group Owner: vpp-dev+ow...@lists.fd.io Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
