...
BugZero found this defect 2863 days ago.
The current implementation of migration_critical_section_concurrency.js uses a failpoint on the MigrationDestinationManager, which however cannot be blocked for more than 30 seconds because this is the timeout for migration commit. This causes build failures on slower machines and debug builds. Instead we should be using the 'hangBeforeLeavingCriticalSection' failpoint which is on the donor shard and can be held indefinitely.
xgen-internal-githook commented on Tue, 11 Apr 2017 20:42:30 +0000: Author: {u'username': u'kaloianm', u'name': u'Kaloian Manassiev', u'email': u'kaloian.manassiev@mongodb.com'} Message: SERVER-28698 Use 'chunkDataCommitted' failpoint in migration_critical_section_concurrency.js (cherry picked from commit 61339fc6cf60938cbc83fd5b6772b10da55d4289) Branch: v3.4 https://github.com/mongodb/mongo/commit/e3650103509da22a0c4dc7c07ab4f9cea9521e8e xgen-internal-githook commented on Mon, 10 Apr 2017 20:06:37 +0000: Author: {u'username': u'kaloianm', u'name': u'Kaloian Manassiev', u'email': u'kaloian.manassiev@mongodb.com'} Message: SERVER-28698 Use 'chunkDataCommitted' failpoint in migration_critical_section_concurrency.js Branch: master https://github.com/mongodb/mongo/commit/61339fc6cf60938cbc83fd5b6772b10da55d4289