commit 2a45cf1abe74ba2a971db9e525abd76134a826d9 Author: Dan Callaghan Date: Mon Aug 8 17:43:31 2016 +1000 Automatic commit of release 23.1 commit bc9e27f789b19b46a47eddfe3b3e167fb0b3a639 Author: Dan Callaghan Date: Mon Aug 8 12:48:27 2016 +1000 docs: add missing bit of blurb for 23 release Change-Id: Iaa96a899bbef7069cbf4f25c68b10c0ac118f9a0 commit 4b967cd72c81fdcb497be3ac386a393970bcdf10 Author: Dan Callaghan Date: Mon Aug 8 11:18:25 2016 +1000 release notes for 23.1 Also added --insecure to the bkr man page, which I just noticed was missing. Change-Id: Id84c215e38d45c548bea9268cb6739851b785701 commit e10081d0d2b91cbbe1acbec4f951f3f8f3bbe1a1 Author: Dan Callaghan Date: Wed Aug 3 16:14:49 2016 +1000 move online data migrations from beaker-init to beakerd Instead of being a manual step which the admin must explicitly invoke after their outage is finished, beakerd will automatically perform the migrations. They are done as part of each iteration of the main scheduler loop, so that it's not running concurrent transactions that will fight for locks. Completed migrations are recorded in the database (similar to how Alembic tracks revisions) so that beakerd knows not to attempt them again. Bug: 1362414 Change-Id: I3810336d851606a8747db3a7e45df1380acaa306 commit 1938a8d0743cb2429cef2040c6214dcdd0b5001f Author: Dan Callaghan Date: Wed Aug 3 17:57:22 2016 +1000 avoid IN with subquery when deleting recipe_task_result rows We need to use MySQL's DELETE... JOIN syntax instead, for decent performance. The IN (SELECT...) version of the query performs a whole table scan which takes many minutes on a large production database, whereas the query using JOIN uses indexes. Bug: 1362439 Change-Id: I69886c07817347fbae43f7cf70ccd2dca44512c5 commit 04a6849ad6e380b80cf3d751c023455cb58db47a Author: Matt Jia Date: Fri Jul 22 16:43:00 2016 +1000 job page: use markdown for rendering recipe whiteboard In the recipe row on the job page, we only show the contents of the first

in the rendered whiteboard to avoid breaking out the layout. Based on a patch contributed by Smitha Ratnam Bug: 1334552 Change-Id: I549b1002700cc8d359327160445c9acc46e6b1ab commit 93bca35a7ad1d6199277cba9728315d2f7b3cb1a Author: mjia@redhat.com Date: Thu Jan 15 12:28:14 2015 +1000 limit request data in all cases in beaker-proxy Change-Id: Ia46ca064a1603d9589abf989b804d217802625bd commit aa72972c7e29e5ab946e3276ef0978ec31e581a2 Author: Dan Callaghan Date: Thu Jul 21 12:18:54 2016 +1000 tests: move fake motd to /usr/share Currently it ends up in /usr/lib/python2.* which means the actual path varies across RHEL6 and RHEL7, which makes life difficult in the dogfood task where we need to configure the right path. Change-Id: I47d938dac5a75e4d44d79941e4dc38fc3a3233b2 commit 4ae73e703a99ba3a31409bce2f3885c2a304df13 Author: Dan Callaghan Date: Thu Jul 21 12:57:20 2016 +1000 test_provision: timeout exception message changed wording in Python 2.7 Currently we get: URLError: but on RHEL7 it is instead: timeout: timed out They both seem fine so let's just have the test case accept either wording. Change-Id: Ia3abb6b826510a34bde85fa3407bd32ae392349c commit edaee1b7e8abf674fd43385bdf1018898653197e Author: Dan Callaghan Date: Thu Jul 21 09:16:41 2016 +1000 tests: explicitly require python-cssselect on RHEL7 Change-Id: I8ad9e4d2b5a49fa0b2e56b2c5fc730692cc14706 commit c6ff333b78555c1dbdf669f48f9a2ac239b34a1f Author: Dan Callaghan Date: Thu Jul 21 13:30:25 2016 +1000 test_database_migration: fix warning about 'waived' column The recipeset.waived column has no SQL-level default so we need to set it explicitly. Fixes this warning: /usr/lib64/python2.7/site-packages/sqlalchemy/engine/default.py:436: Warning: Field 'waived' doesn't have a default value cursor.execute(statement, parameters) Change-Id: Ibfd0a8a400c29244a86a90cec8d553f303cae0d5 commit 032aa53519d7ed2edb79aedbf2a26076083d298c Author: Dan Callaghan Date: Thu Jul 21 13:32:51 2016 +1000 test_database_migration: avoid leaking database connections Strictly speaking, it's beaker-init which is leaking them, although for the real beaker-init command it doesn't really matter because the connections are released when the process exits. In the tests, however, connections could be left lingering with open transactions when the next test is trying to drop the database, which would cause a deadlock. So let's close them properly. Bug: 1356852 Change-Id: I3224724963cf4daf670a50a0b867801b568c8c1c commit afce55b9152815a209816d668a1be0ad9e4f2bee Author: Róman Joost Date: Thu Jul 21 16:32:42 2016 +1000 Remove pointless column_property Change-Id: I9b1ef1c9722d865d37c3cfc1e082914b91f5a8cf commit bf8b234e1ebee2b3b7f2db13edfcbb1d20c08b6b Author: Róman Joost Date: Thu Jul 14 10:55:26 2016 +1000 Error out if ca_cert does not exist If CA_CERT points to a non-existing file, bkr client should not set up a connection with the Hub, but error out. Bug: 1350959 Change-Id: I21a7514d8d65f7a9be295e2b6cfa7e772098a42e commit c3db206cfc903e5a589a4d904872a70693bcb50a Author: Dan Callaghan Date: Tue Jul 12 09:57:12 2016 +1000 fix some errors with pylint 1.5 Change-Id: I32152501a9a94db0610c47e40f840af2bee420a6 commit ec9b511fec1ff38309156b8b126e31976d91e8b1 Author: Dan Callaghan Date: Tue Jul 12 09:56:13 2016 +1000 reduce pylint config to the bare minimum This avoids a bunch of deprecation warnings from newer pylint, complaining that we have set some options that we don't care about. Change-Id: I4fb192bc71ce43787d26828a985526dfe7421422 commit f9ce229d4f72ae84a1eeebf56a21233ca4bab25c Author: Dan Callaghan Date: Fri Jul 8 14:53:29 2016 +1000 /users/+self endpoint to replace auth.who_am_i Bug: 1142566 Change-Id: If7c857fe3da5089e550188a677c92101aef88f08 commit e8db145daa58ce360c473e574e32b248bf9e4cd9 Author: Róman Joost Date: Fri Jul 8 16:28:07 2016 +1000 Clarify how to extend the reservation time. Bug: 1353825 Change-Id: Idce64986c0f46668053a3667323d6478082ed6f1 commit 91927704f0646bf09af88d1c7742266934e31bce Author: Dan Callaghan Date: Mon Jun 13 16:40:59 2016 +1000 bkr --insecure for skipping SSL validity checks Bug: 1345735 Change-Id: Id509ad4c86d35f444f7c1dcd93f4188e16eaf6c3