-
Notifications
You must be signed in to change notification settings - Fork 28.5k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[PySpark] Fix tests with Python 2.6 in 1.0 branch #3668
Conversation
Pyrolite can not unpickle array.array which pickled by Python 2.6, this patch fix it by extend Pyrolite. There is a bug in Pyrolite when unpickle array of float/double, this patch workaround it by reverse the endianness for float/double. This workaround should be removed after Pyrolite have a new release to fix this issue. I had send an PR to Pyrolite to fix it: irmen/Pyrolite#11 Author: Davies Liu <[email protected]> Closes apache#2365 from davies/pickle and squashes the following commits: f44f771 [Davies Liu] enable tests about array 3908f5c [Davies Liu] Merge branch 'master' into pickle c77c87b [Davies Liu] cleanup debugging code 60e4e2f [Davies Liu] support unpickle array.array for Python 2.6 Conflicts: core/src/main/scala/org/apache/spark/api/python/SerDeUtil.scala python/pyspark/tests.py
Test build #24318 has started for PR 3668 at commit
|
Test build #24318 has finished for PR 3668 at commit
|
Test FAILed. |
Looks like we also need to backport #1874 so that the tests can work on Python 2.6? |
@davies would you mind squeezing all of that into this PR so we can fix the build once and for all? |
…2.6 Fixes - Modify python/run-tests to test with Python 2.6 - Use unittest2 when running on Python 2.6. - Fix issue with namedtuple. - Skip TestOutputFormat.test_newhadoop on Python 2.6 until SPARK-2951 is fixed. - Fix MLlib _deserialize_double on Python 2.6. Closes apache#1868. Closes apache#1042. Author: Josh Rosen <[email protected]> Closes apache#1874 from JoshRosen/python2.6 and squashes the following commits: 983d259 [Josh Rosen] [SPARK-2954] Fix MLlib _deserialize_double on Python 2.6. 5d18fd7 [Josh Rosen] [SPARK-2948] [SPARK-2910] [SPARK-2101] Python 2.6 fixes Conflicts: python/pyspark/mllib/_common.py python/pyspark/serializers.py python/pyspark/tests.py
…on 2.6 and unittest2 is not installed ./python/run-tests search a Python 2.6 executable on PATH and use it if available. When using Python 2.6, it is going to import unittest2 module which is not a standard library in Python 2.6, so it fails with ImportError. Author: cocoatomo <[email protected]> Closes apache#2759 from cocoatomo/issues/3867-unittest2-import-error and squashes the following commits: f068eb5 [cocoatomo] [SPARK-3867] ./python/run-tests failed when it run with Python 2.6 and unittest2 is not installed
Test build #24324 has started for PR 3668 at commit
|
@andrewor14 I had added those two commits. |
Test build #24326 has started for PR 3668 at commit
|
Test build #24324 has finished for PR 3668 at commit
|
Test FAILed. |
Test build #24326 has finished for PR 3668 at commit
|
Test FAILed. |
Test build #24337 has started for PR 3668 at commit
|
Test build #24337 has finished for PR 3668 at commit
|
Test FAILed. |
@andrewor14 The failed test may be related to the version of NumPy, I can not reproduce it with 1.8.1 |
Yeah branch 0.9 is also having the same problem. I haven't looked deep into the issue yet but maybe @shaneknapp has a better idea? |
would you guys be adverse to me upping the numpy version on the workers? currently, we're running numpy-1.4.1-9.el6.x86_64, which is ancient compared to the L&G release (1.9.1). anyone know of any backwards-compatibility issues offhand? in the meanwhile, i will set up one of our staging nodes and start testing. i'd rather not to a system-wide update during a release cycle if at all possible... but if i have to do this, i will bring up some new jenkins workers (which i was going to do during tomorrow morning's maintanence anyways) w/an updated numpy specifically for these builds. EDIT: just looked at the spark docs and it looks like i'll indeed have to update numpy. :) |
Yeah I think the case is that in branch 0.9 we still require numpy 1.7 but later branches require only 1.4. For branch-1.0, however, it fails with some other error that I would assume is because we need a more recent version of numpy that is explicitly required. I think we should just upgrade it, since not super likely that we'll be pushing more commits into the current RC. |
ok, sgtm. since we're not blocking the release, i'll just get everything staged and do the upgrade during the scheduled maintenance tomorrow morning. |
Great thanks Shane. |
Jenkins, retest this please. |
Test build #24412 has started for PR 3668 at commit
|
Test build #24412 has finished for PR 3668 at commit
|
Test FAILed. |
Jenkins, retest this please. |
Test build #24414 has started for PR 3668 at commit
|
Test build #24414 has finished for PR 3668 at commit
|
Test PASSed. |
woot! :) On Fri, Dec 12, 2014 at 12:47 PM, UCB AMPLab [email protected]
|
Nice! I'm merging this. |
[SPARK-2951] [PySpark] support unpickle array.array for Python 2.6 Pyrolite can not unpickle array.array which pickled by Python 2.6, this patch fix it by extend Pyrolite. There is a bug in Pyrolite when unpickle array of float/double, this patch workaround it by reverse the endianness for float/double. This workaround should be removed after Pyrolite have a new release to fix this issue. [PySpark] [SPARK-2954] [SPARK-2948] [SPARK-2910] [SPARK-2101] Python 2.6 Fixes - Modify python/run-tests to test with Python 2.6 - Use unittest2 when running on Python 2.6. - Fix issue with namedtuple. - Skip TestOutputFormat.test_newhadoop on Python 2.6 until SPARK-2951 is fixed. - Fix MLlib _deserialize_double on Python 2.6. [SPARK-3867][PySpark] ./python/run-tests failed when it run with Python 2.6 and unittest2 is not installed ./python/run-tests search a Python 2.6 executable on PATH and use it if available. When using Python 2.6, it is going to import unittest2 module which is not a standard library in Python 2.6, so it fails with Import Author: cocoatomo <[email protected]> Author: Josh Rosen <[email protected]> Author: Davies Liu <[email protected]> Author: Davies Liu <[email protected]> Closes #3668 from davies/port_2365 and squashes the following commits: b32583d [Davies Liu] rollback _common.py bda1c72 [cocoatomo] [SPARK-3867][PySpark] ./python/run-tests failed when it run with Python 2.6 and unittest2 is not installed 14ad3d9 [Josh Rosen] [PySpark] [SPARK-2954] [SPARK-2948] [SPARK-2910] [SPARK-2101] Python 2.6 Fixes 7c55cff [Davies Liu] [SPARK-2951] [PySpark] support unpickle array.array for Python 2.6
Do you mind closing this PR since Andrew has merged it into branch-1.0 (9c21316)? GitHub can't auto-close it since it's not against the master branch. |
[SPARK-2951] [PySpark] support unpickle array.array for Python 2.6 Pyrolite can not unpickle array.array which pickled by Python 2.6, this patch fix it by extend Pyrolite. There is a bug in Pyrolite when unpickle array of float/double, this patch workaround it by reverse the endianness for float/double. This workaround should be removed after Pyrolite have a new release to fix this issue. [PySpark] [SPARK-2954] [SPARK-2948] [SPARK-2910] [SPARK-2101] Python 2.6 Fixes - Modify python/run-tests to test with Python 2.6 - Use unittest2 when running on Python 2.6. - Fix issue with namedtuple. - Skip TestOutputFormat.test_newhadoop on Python 2.6 until SPARK-2951 is fixed. - Fix MLlib _deserialize_double on Python 2.6. [SPARK-3867][PySpark] ./python/run-tests failed when it run with Python 2.6 and unittest2 is not installed ./python/run-tests search a Python 2.6 executable on PATH and use it if available. When using Python 2.6, it is going to import unittest2 module which is not a standard library in Python 2.6, so it fails with Import Author: cocoatomo <[email protected]> Author: Josh Rosen <[email protected]> Author: Davies Liu <[email protected]> Author: Davies Liu <[email protected]> Closes apache#3668 from davies/port_2365 and squashes the following commits: b32583d [Davies Liu] rollback _common.py bda1c72 [cocoatomo] [SPARK-3867][PySpark] ./python/run-tests failed when it run with Python 2.6 and unittest2 is not installed 14ad3d9 [Josh Rosen] [PySpark] [SPARK-2954] [SPARK-2948] [SPARK-2910] [SPARK-2101] Python 2.6 Fixes 7c55cff [Davies Liu] [SPARK-2951] [PySpark] support unpickle array.array for Python 2.6 Conflicts: python/pyspark/mllib/tests.py python/pyspark/tests.py python/run-tests
[PySpark] [SPARK-2954] [SPARK-2948] [SPARK-2910] [SPARK-2101] Python 2.6 Fixes - Modify python/run-tests to test with Python 2.6 - Use unittest2 when running on Python 2.6. Author: Josh Rosen <joshrosenapache.org> Closes #3668 from davies/port_2365 and squashes the following commits: Author: cocoatomo <[email protected]> Closes #3968 from davies/fix_python_tests and squashes the following commits: ac4a353 [cocoatomo] [PySpark] Fix tests with Python 2.6 in 1.0 branch
[SPARK-2951] [PySpark] support unpickle array.array for Python 2.6
Pyrolite can not unpickle array.array which pickled by Python 2.6, this patch fix it by extend Pyrolite.
There is a bug in Pyrolite when unpickle array of float/double, this patch workaround it by reverse the endianness for float/double. This workaround should be removed after Pyrolite have a new release to fix this issue.
[PySpark] [SPARK-2954] [SPARK-2948] [SPARK-2910] [SPARK-2101] Python 2.6 Fixes
[SPARK-3867][PySpark] ./python/run-tests failed when it run with Python 2.6 and unittest2 is not installed
./python/run-tests search a Python 2.6 executable on PATH and use it if available.
When using Python 2.6, it is going to import unittest2 module which is not a standard library in Python 2.6, so it fails with Import