summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
...
* SERVER-6008 move '--' out of DAEMON_OPTS rvalueErnie Hershey2013-10-021-1/+1
|
* SERVER-6008 put the logic for DAEMON_OPTS= and -- into the "if which numactl"Aricg2013-10-021-3/+4
| | | | | | I hope this is correct, I have signed the contributor's agreement. Signed-off-by: Ernie Hershey <ernie.hershey@10gen.com>
* SERVER-9907 Allow to skip initial count() in mapreduceRandolph Tan2013-10-021-3/+17
| | | | Do not count total documents to process if filter is given.
* SERVER-9907 Allow to skip initial count() in mapreduceRandolph Tan2013-10-022-5/+15
| | | | | | | Added option to hide the total in the progress meter. Conflicts: src/mongo/util/progress_meter.cpp
* SERVER-9738 mongos should set the CursorNotFound bit when the cursor is not ↵Randolph Tan2013-10-021-10/+2
| | | | | | found Lower invalid cursor log level from 6 to 3.
* SERVER-9738 mongos should set the CursorNotFound bit when the cursor is not ↵Randolph Tan2013-10-021-3/+4
| | | | found
* SERVER-10261 disable ssl session caching (affects java driver)Eric Milkie2013-09-261-8/+2
|
* SERVER-10724 better exception handling in health pollmatt dannenberg2013-09-101-3/+5
|
* SERVER-10596 Globalize formerly per-thread Pool of JS ScopesMathias Stearn2013-08-2710-91/+238
| | | | | | | | | | | | | | | | | | | | | | This ensures that the limit of 10 pooled scopes is actually enforced. With a per-thread Pool, long-lived connections could cause very high memory usage (both real and virtual) even if they haven't used JS in a long time. Manually fixed backport of commit 73841f7a1ec1322d96179eb2712ab438f56add00 Conflicts: jstests/auth/js_scope_leak.js src/mongo/db/auth/auth_external_state.h src/mongo/db/auth/auth_external_state_d.cpp src/mongo/db/auth/auth_external_state_d.h src/mongo/db/auth/auth_external_state_s.cpp src/mongo/db/auth/auth_external_state_s.h src/mongo/db/auth/authorization_session.cpp src/mongo/db/auth/authorization_session.h src/mongo/db/commands/group.cpp src/mongo/db/matcher/expression_where.cpp src/mongo/scripting/engine.cpp
* Add check to prevent future memory leaks like SERVER-10554Mathias Stearn2013-08-232-0/+47
|
* SERVER-10554 Patch memory leak in aggregationMathias Stearn2013-08-232-2/+28
| | | | | | | | | | The issue was that empty documents are represented as a NULL pointer and Value didn't ref count NULL pointers. This is fine for all public interactions with Value, however MutableDocument can change the pointer in-place when updating a nested field. This caused a leak when it transitioned a Value from holding a non-refcounted NULL pointer to holding a real pointer since MutableDocument/MutableValue didn't set the refCount bit.
* post 2.4.6Ernie Hershey2013-08-202-2/+2
|
* BUMP 2.4.6r2.4.6Dan Pasette2013-08-183-3/+3
|
* post 2.4.6-rc1Ernie Hershey2013-08-122-2/+2
|
* BUMP 2.4.6-rc1r2.4.6-rc1Matt Kangas2013-08-102-2/+2
|
* SERVER-10458 sanity check before critical section that all cloned docs sentGreg Studer2013-08-091-3/+30
|
* SERVER-10478 fix batch limit check for _cloneLocs in migrationGreg Studer2013-08-092-3/+67
|
* post 2.4.6-rc0Ernie Hershey2013-08-092-2/+2
|
* missing semicolonr2.4.6-rc0Ernie Hershey2013-08-081-1/+1
|
* BUMP 2.4.6-rc0Ernie Hershey2013-08-082-2/+2
|
* SERVER-10429 disable flaky test until we can fix it properlyEric Milkie2013-08-071-0/+2
|
* SERVER-10437 remove chaining.js testEric Milkie2013-08-071-57/+0
| | | | | | | It is testing functionality that has been deprecated by the new sync source updater. Also, tags.js covers testing chaining with network failures, so there is no further need for this test. chaining.js is racy with respect to the way it activates a failpoint in the socket class, which was causing the test to fail sometimes.
* SERVER-10120 remove2.js buildbot timeout too aggressiveGreg Studer2013-08-051-3/+3
|
* SERVER-8707 Sleep 10 seconds before entering ReplicaSetMonitorWatcher's loopTad Marshall2013-08-041-0/+1
| | | | | | | This restores the startup timing that was changed by commit e3b8349ec305b00bdb7107f2d887549640a15c90, while preserving the smaller time window within the loop between checking for termination and calling ReplicaSetMonitor::checkAll().
* SERVER-10137 sort1.js can't use sh.stopBalancer without a dbGreg Studer2013-08-021-1/+1
|
* SERVER-10135 sort1.js turn off balancer for manual chunk movesGreg Studer2013-08-021-1/+5
|
* SERVER-9934 remove double getmore when pulling new opsEric Milkie2013-08-021-52/+61
| | | | | | | This reduces the potential delay in ending the bgsync thread from 10 seconds to 5 seconds. This is important because assumingPrimaryness waits on this loop before changing state to PRIMARY after an election. To reduce the delay lower than 5 seconds requires further redesign of the code.
* SERVER-9714 buildbot auto2.js don't run connpoolsync before gleGreg Studer2013-08-021-1/+2
|
* SERVER-9913 protect ScopedDbConn constructor in migrate critical sectionGreg Studer2013-08-021-34/+33
|
* SERVER-8899 read-lock while updating ShardChunkManager in trySetVersion()Greg Studer2013-08-021-0/+4
|
* SERVER-10007 compile fail, toString() needed for str::stream()Greg Studer2013-08-021-1/+1
|
* SERVER-10007 better logging for dbclient_rs.cppGreg Studer2013-08-023-20/+136
|
* SERVER-8707 Stop the ReplicaSetMonitorWatcher loop on process terminationTad Marshall2013-08-021-2/+6
| | | | | | | Use StaticObserver to end ReplicaSetMonitorWatcher's loop when a program using the C++ driver exits. Move the sleep to the bottom of the loop to reduce the window between checking for termination and the call to ReplicaSetMonitor::checkAll().
* SERVER-10015 Visual Studio -- add s/config_server_checker_service.{cpp,h}Tad Marshall2013-08-024-0/+16
|
* SERVER-7728 Add more details on why a shard is 'unavailable' for balancingRandolph Tan2013-08-011-2/+12
|
* SERVER-9481 further rollback exception cleanupMatt Dannenberg2013-08-012-23/+13
|
* SERVER-9481 Fixes fluctuating between ROLLBACK and SECONDARY when ROLLBACK ↵Andreas Heck2013-08-012-5/+12
| | | | | | | | | is impossible Only leave ROLLBACK when successful and go to FATAL when ROLLBACK is impossible because we are not willing to rollback so much data Signed-off-by: Matt Dannenberg <matt.dannenberg@10gen.com>
* SERVER-9365 make forced halfway split work with chunks with many docsGreg Studer2013-08-016-10/+147
| | | | Additional test fixes.
* SERVER-10362: call ClientCursor::staticYield to do a pthread_yield rather ↵Dan Pasette2013-08-011-1/+1
| | | | than sleepmicros.
* SERVER-10211 Give stepdown.js more time for the primary to be elected after ↵Spencer T Brody2013-08-011-0/+1
| | | | a reconfig
* SERVER-10362: only use pthread_yield on linux, on osx use sleepmicros(1)Eliot Horowitz2013-08-011-6/+12
|
* SERVER-10362: add option to ClientCursor::staticYield to do a pthread_yield ↵Eliot Horowitz2013-08-012-7/+22
| | | | | | instead of sleepmicros this is mostly because of some platforms (like xen) where sleepmicros is too inaccurate
* SERVER-10313: uassert number conflict fixScott Hernandez2013-08-011-1/+1
|
* SERVER-10313: uassert if null char in js field namesScott Hernandez2013-08-012-0/+16
|
* SERVER-10103 avoid hitting socket exception in the shellEric Milkie2013-08-011-0/+3
|
* SERVER-10103 add debugging outputEric Milkie2013-08-011-1/+3
|
* SERVER-10015 balancer should stop when ConfigServerCheck indicates inconsistencyRandolph Tan2013-08-015-9/+105
| | | | | Conflicts: src/mongo/s/server.cpp
* SERVER-9699: remove clean commandEliot Horowitz2013-08-011-37/+0
| | | | | Conflicts: src/mongo/db/dbcommands_admin.cpp
* More readable SocketException::toString()Mathias Stearn2013-08-011-1/+1
|
* SERVER-9808 Fix issues found in static code analysisAndreas Nilsson2013-07-311-1/+2
|
View Lorry Controller Status