[Impala-ASF-CR] IMPALA-6087: Revisit tests withheld from TPC-DS suite for use of TRUNCATE
Matthew Mulder has posted comments on this change. ( http://gerrit.cloudera.org:8080/8372 ) Change subject: IMPALA-6087: Revisit tests withheld from TPC-DS suite for use of TRUNCATE .. Patch Set 5: (2 comments) http://gerrit.cloudera.org:8080/#/c/8372/5/testdata/workloads/tpcds/queries/tpcds-q26.test File testdata/workloads/tpcds/queries/tpcds-q26.test: http://gerrit.cloudera.org:8080/#/c/8372/5/testdata/workloads/tpcds/queries/tpcds-q26.test@5 PS5, Line 5: 78.33 The TPC-DS spec section 7.5.3.d says "For results from AVG aggregates, the resulting values must be within 1% of the query validation output data when reported to the nearest 1/100th, rounded up." so I guess this is okay. http://gerrit.cloudera.org:8080/#/c/8372/5/testdata/workloads/tpcds/queries/tpcds-q39-1.test File testdata/workloads/tpcds/queries/tpcds-q39-1.test: http://gerrit.cloudera.org:8080/#/c/8372/5/testdata/workloads/tpcds/queries/tpcds-q39-1.test@4 PS5, Line 4: -- ADD ROUND()s TO 4th, 5th, 9th, 10th COLUMNS, USE ACTUAL RESULT AS EXPECTED RESULT. It looks like this is okay per the TPC-DS standard section 4.2.3.4.f.6: Explicit Casting - Scalar functions (such as CAST) whose sole purpose is to affect result precision for operations involving integer columns or values may be applied. The resulting syntax must have equivalent semantic behavior. -- To view, visit http://gerrit.cloudera.org:8080/8372 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I79d2e34621639c8f8c4c4eb0b0944eaefca13a7a Gerrit-Change-Number: 8372 Gerrit-PatchSet: 5 Gerrit-Owner: Tim WoodGerrit-Reviewer: Greg Rahn Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Mostafa Mokhtar Gerrit-Reviewer: Tim Wood Gerrit-Comment-Date: Fri, 10 Nov 2017 03:04:58 + Gerrit-HasComments: Yes
[Impala-ASF-CR] IMPALA-6063: stress test: report test duration
Matthew Mulder has uploaded this change for review. ( http://gerrit.cloudera.org:8080/8339 Change subject: IMPALA-6063: stress test: report test duration .. IMPALA-6063: stress test: report test duration At the end of the stress test, report the number of seconds the test ran like this: Test Duration: 1234 seconds Time spent calculating runtime information isn't counted in the test duration. There are some additional changes to simplify the run_queries method because Flake8 complained that it was too complex and this change was making it even longer. Testing: Ran the stress test and verified the status output and Test Duration output. Change-Id: Ic1769c6c27cf064a330026d12d50562abfaf656f --- M tests/stress/concurrent_select.py 1 file changed, 79 insertions(+), 64 deletions(-) git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/39/8339/1 -- To view, visit http://gerrit.cloudera.org:8080/8339 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: newchange Gerrit-Change-Id: Ic1769c6c27cf064a330026d12d50562abfaf656f Gerrit-Change-Number: 8339 Gerrit-PatchSet: 1 Gerrit-Owner: Matthew Mulder
[Impala-ASF-CR] IMPALA-5376: Implement all TPCDS test cases or alternates for Impala.
Matthew Mulder has posted comments on this change. ( http://gerrit.cloudera.org:8080/8102 ) Change subject: IMPALA-5376: Implement all TPCDS test cases or alternates for Impala. .. Patch Set 12: The stress test is at tests/stress/concurrent_select.py. The list of queries to run is created in load_tpc_queries(). The queries are initially executed in populate_all_queries(). Any query error in populate_all_queries() seems to be fatal. I see two general choices: 1. Do something in load_tpc_queries() to blacklist the unsupported queries. 2. Do something in main() before calling populate_all_queries() to remove the unsupported queries. There are a few options for what "do something" would be. I wouldn't recommend simply skipping queries with errors in populate_all_queries() because that could conceal a regression. -- To view, visit http://gerrit.cloudera.org:8080/8102 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I6e284888600a7a69d1f23fcb7dac21cbb13b7d66 Gerrit-Change-Number: 8102 Gerrit-PatchSet: 12 Gerrit-Owner: Tim WoodGerrit-Reviewer: Alex Behm Gerrit-Reviewer: David Knupp Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Michael Ho Gerrit-Reviewer: Mostafa Mokhtar Gerrit-Reviewer: Tim Wood Gerrit-Comment-Date: Sat, 30 Sep 2017 00:59:02 + Gerrit-HasComments: No
[Impala-ASF-CR] IMPALA-5376: Implement all TPCDS test cases or alternates for Impala.
Matthew Mulder has posted comments on this change. ( http://gerrit.cloudera.org:8080/8102 ) Change subject: IMPALA-5376: Implement all TPCDS test cases or alternates for Impala. .. Patch Set 12: (1 comment) http://gerrit.cloudera.org:8080/#/c/8102/12/testdata/workloads/tpcds/queries/tpcds-q36.test File testdata/workloads/tpcds/queries/tpcds-q36.test: http://gerrit.cloudera.org:8080/#/c/8102/12/testdata/workloads/tpcds/queries/tpcds-q36.test@8 PS12, Line 8: grouping > Impala doesn't seem to accept this syntax. I should have mentioned that this turns out to be a fatal error for the stress test. -- To view, visit http://gerrit.cloudera.org:8080/8102 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I6e284888600a7a69d1f23fcb7dac21cbb13b7d66 Gerrit-Change-Number: 8102 Gerrit-PatchSet: 12 Gerrit-Owner: Tim WoodGerrit-Reviewer: Alex Behm Gerrit-Reviewer: David Knupp Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Michael Ho Gerrit-Reviewer: Mostafa Mokhtar Gerrit-Reviewer: Tim Wood Gerrit-Comment-Date: Sat, 30 Sep 2017 00:16:30 + Gerrit-HasComments: Yes
[Impala-ASF-CR] IMPALA-5376: Implement all TPCDS test cases or alternates for Impala.
Matthew Mulder has posted comments on this change. ( http://gerrit.cloudera.org:8080/8102 ) Change subject: IMPALA-5376: Implement all TPCDS test cases or alternates for Impala. .. Patch Set 12: (1 comment) http://gerrit.cloudera.org:8080/#/c/8102/12/testdata/workloads/tpcds/queries/tpcds-q36.test File testdata/workloads/tpcds/queries/tpcds-q36.test: http://gerrit.cloudera.org:8080/#/c/8102/12/testdata/workloads/tpcds/queries/tpcds-q36.test@8 PS12, Line 8: grouping Impala doesn't seem to accept this syntax. impala.error.HiveServer2Error: AnalysisException: tpcds_300_decimal_parquet.grouping() unknown -- To view, visit http://gerrit.cloudera.org:8080/8102 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I6e284888600a7a69d1f23fcb7dac21cbb13b7d66 Gerrit-Change-Number: 8102 Gerrit-PatchSet: 12 Gerrit-Owner: Tim WoodGerrit-Reviewer: Alex Behm Gerrit-Reviewer: David Knupp Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Michael Ho Gerrit-Reviewer: Mostafa Mokhtar Gerrit-Reviewer: Tim Wood Gerrit-Comment-Date: Sat, 30 Sep 2017 00:15:06 + Gerrit-HasComments: Yes
[Impala-ASF-CR] IMPALA-5376: Implement all TPCDS test cases or alternates for Impala.
Matthew Mulder has posted comments on this change. ( http://gerrit.cloudera.org:8080/8102 ) Change subject: IMPALA-5376: Implement all TPCDS test cases or alternates for Impala. .. Patch Set 11: (1 comment) http://gerrit.cloudera.org:8080/#/c/8102/10/testdata/workloads/tpcds/queries/tpcds-q48.test File testdata/workloads/tpcds/queries/tpcds-q48.test: http://gerrit.cloudera.org:8080/#/c/8102/10/testdata/workloads/tpcds/queries/tpcds-q48.test@1 PS10, Line 1: h stress test fails on this file, probably because this 'h' should be '='. Exception: Expected exactly 1 query to be in file tests/stress/../../testdata/workloads/tpcds/queries/tpcds-q48.test but got 0 -- To view, visit http://gerrit.cloudera.org:8080/8102 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I6e284888600a7a69d1f23fcb7dac21cbb13b7d66 Gerrit-Change-Number: 8102 Gerrit-PatchSet: 11 Gerrit-Owner: Tim WoodGerrit-Reviewer: Alex Behm Gerrit-Reviewer: David Knupp Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Michael Ho Gerrit-Reviewer: Mostafa Mokhtar Gerrit-Reviewer: Tim Wood Gerrit-Comment-Date: Fri, 29 Sep 2017 19:48:47 + Gerrit-HasComments: Yes
[Impala-ASF-CR] Add tpcds-unmodified performance workload.
Matthew Mulder has uploaded this change for review. ( http://gerrit.cloudera.org:8080/5813 Change subject: Add tpcds-unmodified performance workload. .. Add tpcds-unmodified performance workload. Add Mostafa's tpcds-unmodified performance workload for use in the performance test runs. This workload includes 97 queries. Mostafa writes: Before Impala 2.5 we didn't have Runtimefilters, which made TPC-DS queries run very slow, so we ran queries that have explicit partition filters to workaround the limitation. Queries under "tpcds" has those explicit filters, post Impala 2.5 I added the un-modified version of the workload to have more coverage. Q24 and Q67 aren't added because the unmodified version of the query is not supported by Impala and the rewrite version takes a very long time to complete. This workload has already been running nightly for many months out of Mostafa's private branch. This checkin simply does a little cleanup to prepare it for commit to the public repo. How It Was Tested - impala-python -u bin/run-workload.py --client_type=beeswax --workloads=tpcds-unmodified:_1000 --query_iterations 1 --workload_iterations 1 --exec_options='sync_ddl:1;num_scanner_threads:12;MAX_NUM_RUNTIME_FILTERS:50' --impalads=:21000 --num_clients=1 --table_formats=parquet/none --query_names='.*' --results_json_file=/tmp/benchmark_results.json Change-Id: Iea8f3b4e20a30553f1a5a321b2053ff9ca967e92 --- A testdata/workloads/tpcds-unmodified/queries/tpcds-q1.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q10.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q11.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q12.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q13.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q14.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q15.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q16.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q17.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q18.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q19.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q2.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q20.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q21.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q22.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q23.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q25.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q26.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q27.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q28.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q29.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q3.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q30.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q31.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q32.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q33.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q34.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q35.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q36.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q37.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q38.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q39.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q4.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q40.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q41.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q42.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q43.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q44.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q45.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q46.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q47.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q48.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q49.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q5.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q50.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q51.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q52.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q53.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q54.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q55.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q56.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q57.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q58.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q59.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q6.test A testdata/workloads/tpcds-unmodified/queries/tpcds-q60.test A
[Impala-ASF-CR] IMPALA-5376: Implement all TPCDS test cases or alternates for Impala.
Matthew Mulder has posted comments on this change. ( http://gerrit.cloudera.org:8080/8102 ) Change subject: IMPALA-5376: Implement all TPCDS test cases or alternates for Impala. .. Patch Set 10: I found a bug in the stress test. There are already queries named like tpcds-q23-1.test or tpcds-q27a.test that won't be picked up by the stress test because of this regex: file_name_pattern = re.compile(r"%s-%s(q\d+).test$" % (workload, engine)) I'll file a bug (if there isn't one already), but thought I'd share this so you don't wonder why some of your new queries are skipped by the stress test. -- To view, visit http://gerrit.cloudera.org:8080/8102 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I6e284888600a7a69d1f23fcb7dac21cbb13b7d66 Gerrit-Change-Number: 8102 Gerrit-PatchSet: 10 Gerrit-Owner: Tim WoodGerrit-Reviewer: Alex Behm Gerrit-Reviewer: David Knupp Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Michael Ho Gerrit-Reviewer: Mostafa Mokhtar Gerrit-Reviewer: Tim Wood Gerrit-Comment-Date: Fri, 29 Sep 2017 00:19:46 + Gerrit-HasComments: No
[Impala-ASF-CR] IMPALA-5376: Implement all TPCDS test cases or alternates for Impala.
Matthew Mulder has posted comments on this change. ( http://gerrit.cloudera.org:8080/8102 ) Change subject: IMPALA-5376: Implement all TPCDS test cases or alternates for Impala. .. Patch Set 10: Michael commented that the stress test will pick up these new queries. Do you have results from the stress test? -- To view, visit http://gerrit.cloudera.org:8080/8102 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I6e284888600a7a69d1f23fcb7dac21cbb13b7d66 Gerrit-Change-Number: 8102 Gerrit-PatchSet: 10 Gerrit-Owner: Tim WoodGerrit-Reviewer: Alex Behm Gerrit-Reviewer: David Knupp Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Michael Ho Gerrit-Reviewer: Mostafa Mokhtar Gerrit-Reviewer: Tim Wood Gerrit-Comment-Date: Thu, 28 Sep 2017 23:42:28 + Gerrit-HasComments: No
[Impala-ASF-CR] Update stress test with admission control rejected messages
Matthew Mulder has posted comments on this change. Change subject: Update stress test with admission control rejected messages .. Patch Set 2: (1 comment) http://gerrit.cloudera.org:8080/#/c/7854/2/tests/stress/concurrent_select.py File tests/stress/concurrent_select.py: PS2, Line 950: caught_msg == "cancelled": > Either way is okay with me. I guess if we don't take the concurrent_select. As infrequently as the stress test is run on older versions of Impala, I doubt it's worth putting version logic in for this. The least I would do is remove this check, and the most I would do is move this check to its own if block that simply logs a helpful message that this is not supposed to happen after 2.9. -- To view, visit http://gerrit.cloudera.org:8080/7854 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: Ib2f6fa7c0f4e5875fcb92af8f712009ffa02c964 Gerrit-PatchSet: 2 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Tim ArmstrongGerrit-Reviewer: Dan Hecht Gerrit-Reviewer: Matthew Jacobs Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Tim Armstrong Gerrit-HasComments: Yes
[Impala-ASF-CR] Update stress test with admission control rejected messages
Matthew Mulder has posted comments on this change. Change subject: Update stress test with admission control rejected messages .. Patch Set 2: (1 comment) http://gerrit.cloudera.org:8080/#/c/7854/2/tests/stress/concurrent_select.py File tests/stress/concurrent_select.py: PS2, Line 950: caught_msg == "cancelled": > I think also there were other bugs that would cause CANCELLED to overwrite I searched through all of the available stress test results on Jenkins and didn't see this "cancelled" message. -- To view, visit http://gerrit.cloudera.org:8080/7854 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: Ib2f6fa7c0f4e5875fcb92af8f712009ffa02c964 Gerrit-PatchSet: 2 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Tim ArmstrongGerrit-Reviewer: Dan Hecht Gerrit-Reviewer: Matthew Jacobs Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Tim Armstrong Gerrit-HasComments: Yes
[Impala-ASF-CR] IMPALA-5625: stress test: write profile when queries fail
Matthew Mulder has posted comments on this change. Change subject: IMPALA-5625: stress test: write profile when queries fail .. Patch Set 2: (1 comment) http://gerrit.cloudera.org:8080/#/c/7376/2/tests/stress/concurrent_select.py File tests/stress/concurrent_select.py: PS2, Line 1668: get_profile > If you fixed this, can you mark "Done"? Done -- To view, visit http://gerrit.cloudera.org:8080/7376 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: I1dbdf5fcf97d6c5681c9fc8fb9eb448bc459b3b0 Gerrit-PatchSet: 2 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Matthew MulderGerrit-Reviewer: David Knupp Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Tim Wood Gerrit-HasComments: Yes
[Impala-ASF-CR] IMPALA-5625: stress test: write profile when queries fail
Matthew Mulder has uploaded a new patch set (#4). Change subject: IMPALA-5625: stress test: write profile when queries fail .. IMPALA-5625: stress test: write profile when queries fail This change writes query profiles as text files for all of the major query failure reasons in the concurrent_select stress test. 1) Change the --result-hash-log-dir command-line option to --results-dir and update the help text. 2) Introduce two new directories under the directory given by the --results-dir command-line argument: profiles result_hashes 3) Move results into the result_hashes directory. 4) Write the query profile to the profiles directory when a query times out or gets an error or incorrect results. 5) Remove the query profile from the log output for unexpected mem limit exceeded exceptions. Instead, write those to the profiles directory as well. Testing: Ran the stress test with a driver that changes the hashes of some of the query results in the runtime info json file to inject incorrect result failures. Set tight bounds on the mem limit and timeout to ensure there would be timeouts and exceeded memory limit failures. Restarted the NameNode mid test run to induce a query failure. That covers the 4 cases for which an exception is thrown and profile is written for query failures. Verified that the profiles were written for each kind of query failure. Change-Id: I1dbdf5fcf97d6c5681c9fc8fb9eb448bc459b3b0 --- M tests/stress/concurrent_select.py 1 file changed, 92 insertions(+), 60 deletions(-) git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/76/7376/4 -- To view, visit http://gerrit.cloudera.org:8080/7376 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: newpatchset Gerrit-Change-Id: I1dbdf5fcf97d6c5681c9fc8fb9eb448bc459b3b0 Gerrit-PatchSet: 4 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Matthew MulderGerrit-Reviewer: David Knupp Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Tim Wood
[Impala-ASF-CR] get profile() -> fetch and set profile()
Matthew Mulder has abandoned this change. Change subject: get_profile() -> fetch_and_set_profile() .. Abandoned Oops, wrong change id. -- To view, visit http://gerrit.cloudera.org:8080/7804 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: abandon Gerrit-Change-Id: If05d23a4fd5176edce88fd82c85b451812c7d611 Gerrit-PatchSet: 1 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Matthew MulderGerrit-Reviewer: Michael Brown
[Impala-ASF-CR] get profile() -> fetch and set profile()
Matthew Mulder has uploaded a new change for review. http://gerrit.cloudera.org:8080/7804 Change subject: get_profile() -> fetch_and_set_profile() .. get_profile() -> fetch_and_set_profile() Change-Id: If05d23a4fd5176edce88fd82c85b451812c7d611 --- M tests/stress/concurrent_select.py 1 file changed, 5 insertions(+), 5 deletions(-) git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/04/7804/1 -- To view, visit http://gerrit.cloudera.org:8080/7804 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: newchange Gerrit-Change-Id: If05d23a4fd5176edce88fd82c85b451812c7d611 Gerrit-PatchSet: 1 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Matthew Mulder
[Impala-ASF-CR] IMPALA-5625: stress test: write profile when queries fail
Matthew Mulder has posted comments on this change. Change subject: IMPALA-5625: stress test: write profile when queries fail .. Patch Set 2: (1 comment) http://gerrit.cloudera.org:8080/#/c/7376/2/tests/stress/concurrent_select.py File tests/stress/concurrent_select.py: PS2, Line 1668: get_profile > The first word in the docstring is "set". :) fetch_profile() is OK, but the How about fetch_and_set_profile()? -- To view, visit http://gerrit.cloudera.org:8080/7376 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: I1dbdf5fcf97d6c5681c9fc8fb9eb448bc459b3b0 Gerrit-PatchSet: 2 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Matthew MulderGerrit-Reviewer: David Knupp Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Tim Wood Gerrit-HasComments: Yes
[Impala-ASF-CR] IMPALA-5625: stress test: write profile when queries fail
Matthew Mulder has uploaded a new patch set (#3). Change subject: IMPALA-5625: stress test: write profile when queries fail .. IMPALA-5625: stress test: write profile when queries fail This change writes query profiles as text files for all of the major query failure reasons in the concurrent_select stress test. 1) Change the --result-hash-log-dir command-line option to --results-dir and update the help text. 2) Introduce two new directories under the directory given by the --results-dir command-line argument: profiles result_hashes 3) Move results into the result_hashes directory. 4) Write the query profile to the profiles directory when a query times out or gets an error or incorrect results. 5) Remove the query profile from the log output for unexpected mem limit exceeded exceptions. Instead, write those to the profiles directory as well. Testing: Ran the stress test with a driver that changes the hashes of some of the query results in the runtime info json file to inject incorrect result failures. Set tight bounds on the mem limit and timeout to ensure there would be timeouts and exceeded memory limit failures. Restarted the NameNode mid test run to induce a query failure. That covers the 4 cases for which an exception is thrown and profile is written for query failures. Verified that the profiles were written for each kind of query failure. Change-Id: I1dbdf5fcf97d6c5681c9fc8fb9eb448bc459b3b0 --- M tests/stress/concurrent_select.py 1 file changed, 92 insertions(+), 60 deletions(-) git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/76/7376/3 -- To view, visit http://gerrit.cloudera.org:8080/7376 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: newpatchset Gerrit-Change-Id: I1dbdf5fcf97d6c5681c9fc8fb9eb448bc459b3b0 Gerrit-PatchSet: 3 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Matthew MulderGerrit-Reviewer: David Knupp Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Tim Wood
[Impala-ASF-CR] IMPALA-5625: write profile when query times out
Matthew Mulder has posted comments on this change. Change subject: IMPALA-5625: write profile when query times out .. Patch Set 1: (8 comments) > Matt, you uploaded a new patch set, but please reply to the inline > comments with either "done" or some other explanation. I replied to them with "Done", but they show up as Draft, so I wonder if you don't see them. http://gerrit.cloudera.org:8080/#/c/7376/1//COMMIT_MSG Commit Message: Line 22: > We typically have a testing done section. Done http://gerrit.cloudera.org:8080/#/c/7376/1/tests/stress/concurrent_select.py File tests/stress/concurrent_select.py: PS1, Line 691: "{0}" > Please add query ID: to the string here. Done Line 699: raise Exception( > Can you also add support for saving profiles when incorrect results occur? Done PS1, Line 704: {0} > Please add "query ID" to the string. Done PS1, Line 748: if not (report.profile and report.query_id): : return > Do you know often this path is executed? This is executed every time the profile is attempted to be written because of a query failure. This check cowardly refuses to write the profile if there is no profile or if we fail to get the query id. This could happen if there's a communication error or some other serious problem. Line 865: LOG.debug("Query id is %s", report.query_id) > Should the comma be a % sign? The prior version uses a comma, and the test output looks correct. It looks like all of the LOG.debug function calls use a comma. It would probably work to use a %, but it's better to follow the current style without good cause to change. Line 871: report.profile = cursor.get_profile() > Do we want to set should_cancel here since time is exceeded? should_cancel indicates whether the query timeout was purposely shortened to induce a cancelation as part of the stress procedure. See line 642. Looks like I should have documented this in the function doc. PS1, Line 1704: "--result-hash-log-dir" > Now that this contains profiles, too, should this be renamed? Yeah, good point. I should update the help text as well. How about "--results-dir" which is slightly more generic, or "--artifacts-dir"? I don't want to be too specific with something like "--results-and-profiles-dir" because then it would have to be changed again when a new kind of artifact, say CSV files or charts, are added. Done -- To view, visit http://gerrit.cloudera.org:8080/7376 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: I1dbdf5fcf97d6c5681c9fc8fb9eb448bc459b3b0 Gerrit-PatchSet: 1 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Matthew MulderGerrit-Reviewer: David Knupp Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Tim Wood Gerrit-HasComments: Yes
[Impala-ASF-CR] IMPALA-5625: write profile when query times out
Matthew Mulder has uploaded a new patch set (#2). Change subject: IMPALA-5625: write profile when query times out .. IMPALA-5625: write profile when query times out This change writes query profiles as text files for all of the major query failure reasons in the concurrent_select stress test. 1) Change the --result-hash-log-dir command-line option to --results-dir and update the help text. 2) Introduce two new directories under the directory given by the --results-dir command-line argument: profiles result_hashes 3) Move results into the result_hashes directory. 4) Write the query profile to the profiles directory when a query times out or gets an error or incorrect results. 5) Remove the query profile from the log output for unexpected mem limit exceeded exceptions. Instead, write those to the the profiles directory as well. Testing: Ran the stress test with a driver that changes the hashes of some of the query results in the runtime info json file to inject incorrect result failures. Set tight bounds on the mem limit and timeout to ensure there would be timeouts and exceeded memory limit failures. Restarted the NameNode mid test run to induce a query failure. That covers the 4 cases for which an exception is thrown and profile is written for query failures. Verified that the profiles were written for each kind of query failure. Change-Id: I1dbdf5fcf97d6c5681c9fc8fb9eb448bc459b3b0 --- M tests/stress/concurrent_select.py 1 file changed, 90 insertions(+), 60 deletions(-) git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/76/7376/2 -- To view, visit http://gerrit.cloudera.org:8080/7376 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: newpatchset Gerrit-Change-Id: I1dbdf5fcf97d6c5681c9fc8fb9eb448bc459b3b0 Gerrit-PatchSet: 2 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Matthew MulderGerrit-Reviewer: David Knupp Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Tim Wood
[Impala-ASF-CR] IMPALA-5625: write profile when query times out
Matthew Mulder has uploaded a new change for review. http://gerrit.cloudera.org:8080/7376 Change subject: IMPALA-5625: write profile when query times out .. IMPALA-5625: write profile when query times out Introduces two new directories under the directory given by the --result-hash-log-dir command-line argument: profiles results Move results into the results directory. When a query times out or gets an error, write the query profile to the profiles directory. Remove the query profile from the log output for unexpected mem limit exceeded exceptions. Instead, write those to the the profiles directory as well. Change-Id: I1dbdf5fcf97d6c5681c9fc8fb9eb448bc459b3b0 --- M tests/stress/concurrent_select.py 1 file changed, 48 insertions(+), 33 deletions(-) git pull ssh://gerrit.cloudera.org:29418/Impala-ASF refs/changes/76/7376/1 -- To view, visit http://gerrit.cloudera.org:8080/7376 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: newchange Gerrit-Change-Id: I1dbdf5fcf97d6c5681c9fc8fb9eb448bc459b3b0 Gerrit-PatchSet: 1 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Matthew Mulder
[Impala-ASF-CR] IMPALA-5281: stress test: introduce stricter pass guidelines
Matthew Mulder has posted comments on this change. Change subject: IMPALA-5281: stress test: introduce stricter pass guidelines .. Patch Set 1: Code-Review+1 I reviewed the code and performed a test run, but I'm relying on Michael's testing of changing the result hashes to produce incorrect results. -- To view, visit http://gerrit.cloudera.org:8080/7282 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: I9f2174a527193ae01be45b8ed56315c465883346 Gerrit-PatchSet: 1 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Michael BrownGerrit-Reviewer: David Knupp Gerrit-Reviewer: Matthew Mulder Gerrit-HasComments: No
[Impala-ASF-CR] IMPALA-5263: test infra: support CA bundles with secure clusters
Matthew Mulder has posted comments on this change. Change subject: IMPALA-5263: test infra: support CA bundles with secure clusters .. Patch Set 2: The code looks good to me. Do you have a Jenkins run that uses this code? -- To view, visit http://gerrit.cloudera.org:8080/7152 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: Ifb9e466e4b7cde704cdc4cf98159c068c0a400a9 Gerrit-PatchSet: 2 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Michael BrownGerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Tim Wood Gerrit-HasComments: No
[Impala-ASF-CR] IMPALA-5162,IMPALA-5163: stress test support on secure clusters
Matthew Mulder has posted comments on this change. Change subject: IMPALA-5162,IMPALA-5163: stress test support on secure clusters .. Patch Set 1: Code-Review+1 -- To view, visit http://gerrit.cloudera.org:8080/6763 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: I0daad57bb8ceeb5071b75125f11c1997ed7e0179 Gerrit-PatchSet: 1 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Michael BrownGerrit-Reviewer: Alex Behm Gerrit-Reviewer: David Knupp Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Michael Brown Gerrit-Reviewer: Mostafa Mokhtar Gerrit-HasComments: No
[Impala-ASF-CR] IMPALA-5162,IMPALA-5163: stress test support on secure clusters
Matthew Mulder has posted comments on this change. Change subject: IMPALA-5162,IMPALA-5163: stress test support on secure clusters .. Patch Set 1: (1 comment) http://gerrit.cloudera.org:8080/#/c/6763/1/tests/comparison/cluster.py File tests/comparison/cluster.py: PS1, Line 404: Why is this removed? -- To view, visit http://gerrit.cloudera.org:8080/6763 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: I0daad57bb8ceeb5071b75125f11c1997ed7e0179 Gerrit-PatchSet: 1 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Michael BrownGerrit-Reviewer: David Knupp Gerrit-Reviewer: Matthew Mulder Gerrit-Reviewer: Mostafa Mokhtar Gerrit-HasComments: Yes