Re: Review Request 25972: Make thermos more cognizant of user deletions
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/#review55537 --- thanks, build-support/jenkins/build.sh passed, going to merge to master. - Brian Wickman On Oct. 3, 2014, 10:16 p.m., Brian Wickman wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/ --- (Updated Oct. 3, 2014, 10:16 p.m.) Review request for Aurora, Joe Smith and Kevin Sweeney. Bugs: AURORA-175 https://issues.apache.org/jira/browse/AURORA-175 Repository: aurora Description --- This is changes 1 of 2 for AURORA-175. The second change adds an exit status contract between Aurora and Thermos. This change allows a process in the Thermos state machine to go directly from WAITING - FAILED if the user associated with that process has been deleted. It also persists a 'uid' field to the Thermos RunnerHeader so that we can check against process UIDs which in theory have higher fidelity than usernames. Diffs - src/main/python/apache/thermos/common/ckpt.py 7df179b135e896d52655667f0707850a4bc33068 src/main/python/apache/thermos/core/helper.py 82c68f3af424d3309db3816080465edd1ff1a87c src/main/python/apache/thermos/core/process.py 4889e636c2fa7325852684cecd87a2123714144d src/main/python/apache/thermos/core/runner.py 31f40713ec32c0626566899caec76f2a9985c7bf src/main/thrift/org/apache/thermos/thermos_internal.thrift 0cea1057230b367c9d55b773854453cee2f9fce0 src/test/python/apache/thermos/core/test_helper.py 53f1e467283b45598b3f1865951c42b13579a512 src/test/python/apache/thermos/core/test_process.py 223393d62d85dbcd3da3c2ca0d5558e8209aa1d1 Diff: https://reviews.apache.org/r/25972/diff/ Testing --- Thanks, Brian Wickman
Re: Review Request 25972: Make thermos more cognizant of user deletions
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/ --- (Updated Oct. 3, 2014, 5:55 p.m.) Review request for Aurora, Joe Smith and Kevin Sweeney. Changes --- Address Kevin's feedback Bugs: AURORA-175 https://issues.apache.org/jira/browse/AURORA-175 Repository: aurora Description --- This is changes 1 of 2 for AURORA-175. The second change adds an exit status contract between Aurora and Thermos. This change allows a process in the Thermos state machine to go directly from WAITING - FAILED if the user associated with that process has been deleted. It also persists a 'uid' field to the Thermos RunnerHeader so that we can check against process UIDs which in theory have higher fidelity than usernames. Diffs (updated) - src/main/python/apache/thermos/common/ckpt.py 7df179b135e896d52655667f0707850a4bc33068 src/main/python/apache/thermos/core/helper.py 82c68f3af424d3309db3816080465edd1ff1a87c src/main/python/apache/thermos/core/process.py 4889e636c2fa7325852684cecd87a2123714144d src/main/python/apache/thermos/core/runner.py 31f40713ec32c0626566899caec76f2a9985c7bf src/main/thrift/org/apache/thermos/thermos_internal.thrift 0cea1057230b367c9d55b773854453cee2f9fce0 src/test/python/apache/thermos/core/test_helper.py 53f1e467283b45598b3f1865951c42b13579a512 src/test/python/apache/thermos/core/test_process.py 223393d62d85dbcd3da3c2ca0d5558e8209aa1d1 Diff: https://reviews.apache.org/r/25972/diff/ Testing --- Thanks, Brian Wickman
Re: Review Request 25972: Make thermos more cognizant of user deletions
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/#review55371 --- Looks great- thanks! src/test/python/apache/thermos/core/test_helper.py https://reviews.apache.org/r/25972/#comment95731 from psutil._common import puids - Joe Smith On Oct. 3, 2014, 10:55 a.m., Brian Wickman wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/ --- (Updated Oct. 3, 2014, 10:55 a.m.) Review request for Aurora, Joe Smith and Kevin Sweeney. Bugs: AURORA-175 https://issues.apache.org/jira/browse/AURORA-175 Repository: aurora Description --- This is changes 1 of 2 for AURORA-175. The second change adds an exit status contract between Aurora and Thermos. This change allows a process in the Thermos state machine to go directly from WAITING - FAILED if the user associated with that process has been deleted. It also persists a 'uid' field to the Thermos RunnerHeader so that we can check against process UIDs which in theory have higher fidelity than usernames. Diffs - src/main/python/apache/thermos/common/ckpt.py 7df179b135e896d52655667f0707850a4bc33068 src/main/python/apache/thermos/core/helper.py 82c68f3af424d3309db3816080465edd1ff1a87c src/main/python/apache/thermos/core/process.py 4889e636c2fa7325852684cecd87a2123714144d src/main/python/apache/thermos/core/runner.py 31f40713ec32c0626566899caec76f2a9985c7bf src/main/thrift/org/apache/thermos/thermos_internal.thrift 0cea1057230b367c9d55b773854453cee2f9fce0 src/test/python/apache/thermos/core/test_helper.py 53f1e467283b45598b3f1865951c42b13579a512 src/test/python/apache/thermos/core/test_process.py 223393d62d85dbcd3da3c2ca0d5558e8209aa1d1 Diff: https://reviews.apache.org/r/25972/diff/ Testing --- Thanks, Brian Wickman
Re: Review Request 25972: Make thermos more cognizant of user deletions
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/#review55370 --- Ship it! src/main/python/apache/thermos/core/helper.py https://reviews.apache.org/r/25972/#comment95730 Should we log.debug on the absence of uid from the checkpoint record? - Kevin Sweeney On Oct. 3, 2014, 10:55 a.m., Brian Wickman wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/ --- (Updated Oct. 3, 2014, 10:55 a.m.) Review request for Aurora, Joe Smith and Kevin Sweeney. Bugs: AURORA-175 https://issues.apache.org/jira/browse/AURORA-175 Repository: aurora Description --- This is changes 1 of 2 for AURORA-175. The second change adds an exit status contract between Aurora and Thermos. This change allows a process in the Thermos state machine to go directly from WAITING - FAILED if the user associated with that process has been deleted. It also persists a 'uid' field to the Thermos RunnerHeader so that we can check against process UIDs which in theory have higher fidelity than usernames. Diffs - src/main/python/apache/thermos/common/ckpt.py 7df179b135e896d52655667f0707850a4bc33068 src/main/python/apache/thermos/core/helper.py 82c68f3af424d3309db3816080465edd1ff1a87c src/main/python/apache/thermos/core/process.py 4889e636c2fa7325852684cecd87a2123714144d src/main/python/apache/thermos/core/runner.py 31f40713ec32c0626566899caec76f2a9985c7bf src/main/thrift/org/apache/thermos/thermos_internal.thrift 0cea1057230b367c9d55b773854453cee2f9fce0 src/test/python/apache/thermos/core/test_helper.py 53f1e467283b45598b3f1865951c42b13579a512 src/test/python/apache/thermos/core/test_process.py 223393d62d85dbcd3da3c2ca0d5558e8209aa1d1 Diff: https://reviews.apache.org/r/25972/diff/ Testing --- Thanks, Brian Wickman
Re: Review Request 25972: Make thermos more cognizant of user deletions
On Oct. 3, 2014, 6:17 p.m., Joe Smith wrote: src/test/python/apache/thermos/core/test_helper.py, line 33 https://reviews.apache.org/r/25972/diff/2/?file=713532#file713532line33 from psutil._common import puids Looks like I wasn't using this at all, so dropped entirely. - Brian --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/#review55371 --- On Oct. 3, 2014, 5:55 p.m., Brian Wickman wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/ --- (Updated Oct. 3, 2014, 5:55 p.m.) Review request for Aurora, Joe Smith and Kevin Sweeney. Bugs: AURORA-175 https://issues.apache.org/jira/browse/AURORA-175 Repository: aurora Description --- This is changes 1 of 2 for AURORA-175. The second change adds an exit status contract between Aurora and Thermos. This change allows a process in the Thermos state machine to go directly from WAITING - FAILED if the user associated with that process has been deleted. It also persists a 'uid' field to the Thermos RunnerHeader so that we can check against process UIDs which in theory have higher fidelity than usernames. Diffs - src/main/python/apache/thermos/common/ckpt.py 7df179b135e896d52655667f0707850a4bc33068 src/main/python/apache/thermos/core/helper.py 82c68f3af424d3309db3816080465edd1ff1a87c src/main/python/apache/thermos/core/process.py 4889e636c2fa7325852684cecd87a2123714144d src/main/python/apache/thermos/core/runner.py 31f40713ec32c0626566899caec76f2a9985c7bf src/main/thrift/org/apache/thermos/thermos_internal.thrift 0cea1057230b367c9d55b773854453cee2f9fce0 src/test/python/apache/thermos/core/test_helper.py 53f1e467283b45598b3f1865951c42b13579a512 src/test/python/apache/thermos/core/test_process.py 223393d62d85dbcd3da3c2ca0d5558e8209aa1d1 Diff: https://reviews.apache.org/r/25972/diff/ Testing --- Thanks, Brian Wickman
Re: Review Request 25972: Make thermos more cognizant of user deletions
On Oct. 3, 2014, 6:18 p.m., Kevin Sweeney wrote: src/main/python/apache/thermos/core/helper.py, line 157 https://reviews.apache.org/r/25972/diff/2/?file=713528#file713528line157 Should we log.debug on the absence of uid from the checkpoint record? Sure, won't hurt. - Brian --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/#review55370 --- On Oct. 3, 2014, 5:55 p.m., Brian Wickman wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/ --- (Updated Oct. 3, 2014, 5:55 p.m.) Review request for Aurora, Joe Smith and Kevin Sweeney. Bugs: AURORA-175 https://issues.apache.org/jira/browse/AURORA-175 Repository: aurora Description --- This is changes 1 of 2 for AURORA-175. The second change adds an exit status contract between Aurora and Thermos. This change allows a process in the Thermos state machine to go directly from WAITING - FAILED if the user associated with that process has been deleted. It also persists a 'uid' field to the Thermos RunnerHeader so that we can check against process UIDs which in theory have higher fidelity than usernames. Diffs - src/main/python/apache/thermos/common/ckpt.py 7df179b135e896d52655667f0707850a4bc33068 src/main/python/apache/thermos/core/helper.py 82c68f3af424d3309db3816080465edd1ff1a87c src/main/python/apache/thermos/core/process.py 4889e636c2fa7325852684cecd87a2123714144d src/main/python/apache/thermos/core/runner.py 31f40713ec32c0626566899caec76f2a9985c7bf src/main/thrift/org/apache/thermos/thermos_internal.thrift 0cea1057230b367c9d55b773854453cee2f9fce0 src/test/python/apache/thermos/core/test_helper.py 53f1e467283b45598b3f1865951c42b13579a512 src/test/python/apache/thermos/core/test_process.py 223393d62d85dbcd3da3c2ca0d5558e8209aa1d1 Diff: https://reviews.apache.org/r/25972/diff/ Testing --- Thanks, Brian Wickman
Re: Review Request 25972: Make thermos more cognizant of user deletions
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/ --- (Updated Oct. 3, 2014, 10:16 p.m.) Review request for Aurora, Joe Smith and Kevin Sweeney. Changes --- Address feedback. Bugs: AURORA-175 https://issues.apache.org/jira/browse/AURORA-175 Repository: aurora Description --- This is changes 1 of 2 for AURORA-175. The second change adds an exit status contract between Aurora and Thermos. This change allows a process in the Thermos state machine to go directly from WAITING - FAILED if the user associated with that process has been deleted. It also persists a 'uid' field to the Thermos RunnerHeader so that we can check against process UIDs which in theory have higher fidelity than usernames. Diffs (updated) - src/main/python/apache/thermos/common/ckpt.py 7df179b135e896d52655667f0707850a4bc33068 src/main/python/apache/thermos/core/helper.py 82c68f3af424d3309db3816080465edd1ff1a87c src/main/python/apache/thermos/core/process.py 4889e636c2fa7325852684cecd87a2123714144d src/main/python/apache/thermos/core/runner.py 31f40713ec32c0626566899caec76f2a9985c7bf src/main/thrift/org/apache/thermos/thermos_internal.thrift 0cea1057230b367c9d55b773854453cee2f9fce0 src/test/python/apache/thermos/core/test_helper.py 53f1e467283b45598b3f1865951c42b13579a512 src/test/python/apache/thermos/core/test_process.py 223393d62d85dbcd3da3c2ca0d5558e8209aa1d1 Diff: https://reviews.apache.org/r/25972/diff/ Testing --- Thanks, Brian Wickman
Re: Review Request 25972: Make thermos more cognizant of user deletions
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/#review55410 --- Ship it! Ship It! - Joe Smith On Oct. 3, 2014, 3:16 p.m., Brian Wickman wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/ --- (Updated Oct. 3, 2014, 3:16 p.m.) Review request for Aurora, Joe Smith and Kevin Sweeney. Bugs: AURORA-175 https://issues.apache.org/jira/browse/AURORA-175 Repository: aurora Description --- This is changes 1 of 2 for AURORA-175. The second change adds an exit status contract between Aurora and Thermos. This change allows a process in the Thermos state machine to go directly from WAITING - FAILED if the user associated with that process has been deleted. It also persists a 'uid' field to the Thermos RunnerHeader so that we can check against process UIDs which in theory have higher fidelity than usernames. Diffs - src/main/python/apache/thermos/common/ckpt.py 7df179b135e896d52655667f0707850a4bc33068 src/main/python/apache/thermos/core/helper.py 82c68f3af424d3309db3816080465edd1ff1a87c src/main/python/apache/thermos/core/process.py 4889e636c2fa7325852684cecd87a2123714144d src/main/python/apache/thermos/core/runner.py 31f40713ec32c0626566899caec76f2a9985c7bf src/main/thrift/org/apache/thermos/thermos_internal.thrift 0cea1057230b367c9d55b773854453cee2f9fce0 src/test/python/apache/thermos/core/test_helper.py 53f1e467283b45598b3f1865951c42b13579a512 src/test/python/apache/thermos/core/test_process.py 223393d62d85dbcd3da3c2ca0d5558e8209aa1d1 Diff: https://reviews.apache.org/r/25972/diff/ Testing --- Thanks, Brian Wickman
Re: Review Request 25972: Make thermos more cognizant of user deletions
--- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/#review54470 --- src/main/python/apache/thermos/core/helper.py https://reviews.apache.org/r/25972/#comment94614 nullable as well? src/main/python/apache/thermos/core/helper.py https://reviews.apache.org/r/25972/#comment94616 Logic is inverted - this log message belongs below src/test/python/apache/thermos/core/test_helper.py https://reviews.apache.org/r/25972/#comment94618 why does this need to be a contextmanager? - Kevin Sweeney On Sept. 23, 2014, 5:58 p.m., Brian Wickman wrote: --- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/25972/ --- (Updated Sept. 23, 2014, 5:58 p.m.) Review request for Aurora, Joe Smith and Kevin Sweeney. Bugs: AURORA-175 https://issues.apache.org/jira/browse/AURORA-175 Repository: aurora Description --- This is changes 1 of 2 for AURORA-175. The second change adds an exit status contract between Aurora and Thermos. This change allows a process in the Thermos state machine to go directly from WAITING - FAILED if the user associated with that process has been deleted. It also persists a 'uid' field to the Thermos RunnerHeader so that we can check against process UIDs which in theory have higher fidelity than usernames. Diffs - src/main/python/apache/thermos/common/ckpt.py 7df179b135e896d52655667f0707850a4bc33068 src/main/python/apache/thermos/core/helper.py 82c68f3af424d3309db3816080465edd1ff1a87c src/main/python/apache/thermos/core/process.py 4889e636c2fa7325852684cecd87a2123714144d src/main/python/apache/thermos/core/runner.py 31f40713ec32c0626566899caec76f2a9985c7bf src/main/thrift/org/apache/thermos/thermos_internal.thrift 0cea1057230b367c9d55b773854453cee2f9fce0 src/test/python/apache/thermos/core/test_helper.py 53f1e467283b45598b3f1865951c42b13579a512 src/test/python/apache/thermos/core/test_process.py 223393d62d85dbcd3da3c2ca0d5558e8209aa1d1 Diff: https://reviews.apache.org/r/25972/diff/ Testing --- Thanks, Brian Wickman