summaryrefslogtreecommitdiff
path: root/src
Commit message (Collapse)AuthorAgeFilesLines
* SERVER-14081 Add more deletedList buckets to cover more sizesMathias Stearn2014-10-1713-84/+370
| | | | | | | This is done in a backward compatible way without orphaning any old data. Downgrading after this patch *will* orphan large (>=4MB) deleted records, but they will be usable again after upgrading. Additionally, the compact command should not be run on a downgraded server until SERVER-15319 is resolved.
* SERVER-15273 Only quantize to exact bucket sizesMathias Stearn2014-10-1711-802/+132
| | | | Prep for new freelist implementation.
* SERVER-13635 release record store before restarting engine in Restart1 test caseBenety Goh2014-10-171-19/+23
|
* SERVER-14352 (rocksdb) fixed getSpaceUsedBytes()Igor Canadi2014-10-173-25/+25
| | | | Signed-off-by: Benety Goh <benety@mongodb.com>
* SERVER-15535 Do not consider a remote node electable if it explicitly ↵Andy Schwerin2014-10-173-1/+13
| | | | reports that it is not.
* SERVER-15535 Store a heartbeat response in MemberHeartbeatData.Andy Schwerin2014-10-173-79/+59
|
* SERVER-15535 Remove _configIndex field from MemberHeartbeatData structure.Andy Schwerin2014-10-173-78/+62
| | | | Also, do not seUpValues on self, as they are never consulted.
* SERVER-15705 (rocksdb) fixed numRecords and numEntries to work with recovery ↵sdong2014-10-179-60/+122
| | | | | | unit Signed-off-by: Benety Goh <benety@mongodb.com>
* SERVER-15576 Floating-point inaccuracy when unhashing GeoHash to boxSiyuan Zhou2014-10-177-44/+360
|
* SERVER-14702: add syncingTo to self output of replSetGetStatusScott Hernandez2014-10-172-1/+11
|
* SERVER-14982: don't take lock to schedule maint modeScott Hernandez2014-10-172-1/+7
|
* SERVER-15535: report info about unknown nodes in replSetGetStatusScott Hernandez2014-10-173-74/+70
|
* SERVER-15700 SERVER-15701 fixed compilation errorBenety Goh2014-10-174-4/+5
|
* SERVER-15702 During collection scans, read the document from theJohn Esmet2014-10-171-1/+1
| | | | | | | | | RecordIterator instead of the Collection. This way the RecordIterator class can optimize document fetch for disklocs that were just returned. Closes #829 Signed-off-by: Benety Goh <benety@mongodb.com>
* SERVER-15701 Pass the old object down to updateWithDamages in the formJohn Esmet2014-10-1714-28/+44
| | | | | | | | | of a RecordData. This way, the storage engine does not need to re-read the old value in order to perform the update. Closes #828 Signed-off-by: Benety Goh <benety@mongodb.com>
* SERVER-15700 Modify SortedDataInterface to not require unindex() to returnJohn Esmet2014-10-179-38/+43
| | | | | | | | | whether the index row was deleted. Fix unit tests to manually verify that unindex() operations had the desired effect (in the same operation context) Closes #827 Signed-off-by: Benety Goh <benety@mongodb.com>
* SERVER-6062 clean up of str::stripTrailing. added unit test.Michael Ivanov2014-10-174-6/+100
| | | | | | Closes #795 Signed-off-by: Benety Goh <benety@mongodb.com>
* SERVER-13635 (rocksdb) pass whether an index is unique or not down through ↵Benety Goh2014-10-173-3/+5
| | | | SortedDataInterface
* SERVER-13635: OperationContext doesn't always have a RecoveryUnitEliot Horowitz2014-10-171-2/+4
|
* fix txn read boundaries and lockingEliot Horowitz2014-10-171-7/+10
|
* SERVER-15683: fix caching of Client when there is noneEliot Horowitz2014-10-171-1/+3
|
* SERVER-13635 take ownership of shard key in shardCollectionGreg Studer2014-10-161-2/+4
|
* SERVER-15486 LogComponentVerbositySetting validationMatt Kangas2014-10-161-54/+80
| | | | | | | | Implement _set via parseLogComponentSettings. Also extract most of LogComponentVerbositySetting.append to new method _get. Remove need to prune mutablebson doc produced by _get. Replace typedefs with 'using'.
* SERVER-15486 parse_log_component_settingsMatt Kangas2014-10-165-0/+377
|
* SERVER-13635 fixed createCollection shell helper to understand storageEngine ↵Benety Goh2014-10-161-0/+16
| | | | option
* SERVER-14973 consolidate shard key parsing, cleanup shard key patternsGreg Studer2014-10-1650-1285/+2871
|
* SERVER-13635: add hook to RecoveryUnit for when we are take out of and put ↵Eliot Horowitz2014-10-162-0/+5
| | | | back into service
* SERVER-13635: fix race in kv db catalog entryEliot Horowitz2014-10-161-1/+2
|
* SERVER-13635: txn boundary for chunk migration getting disk locsEliot Horowitz2014-10-161-0/+1
|
* SERVER-15683: add command to introspect OperationContext and RecoveryUnitEliot Horowitz2014-10-165-22/+75
|
* SERVER-13635: better txn lines for replication syncEliot Horowitz2014-10-165-14/+36
|
* SERVER-13635: pass whether an index is unique or not down through ↵Eliot Horowitz2014-10-1625-221/+236
| | | | SortedDataInterface
* SERVER-15535 Do not immediately stand for reelection after stepping down for ↵Andy Schwerin2014-10-161-0/+6
| | | | a higher priority node.
* SERVER-15535 adjust resync command handling synchronizationEric Milkie2014-10-161-1/+9
|
* SERVER-14668: Use server parameter to enable CLLGeert Bosch2014-10-162-7/+9
| | | | This does not affect storage engines that support document level locking.
* SERVER-15244 More aggressive auto-split when total chunks are smallRandolph Tan2014-10-162-10/+34
|
* SERVER-15647: make Extent Manager thread-safe and enable MMAPv1 CLLGeert Bosch2014-10-1610-82/+176
|
* SERVER-15535 update slavemap optimes on heartbeatsEric Milkie2014-10-162-0/+39
|
* SERVER-15535 revert freshness changes w.r.t. priorityEric Milkie2014-10-161-14/+1
|
* SERVER-15535 Make all calls to exitCleanly but the first hang forever.Andy Schwerin2014-10-161-2/+6
| | | | | | | | Some components, notably the replication coordinator, cannot be shutdown repeatedly, but also may not be shutdown while holding the global exclusive lock. With this change, only one call to exitCleanly will proceed to completion; all others will spin in a long sleep loop, neither returning nor engaging in shut down tasks.
* SERVER-15639 Correctly handle >1 pred on same text index prefix fieldJason Rassi2014-10-162-2/+15
|
* SERVER-15336: Reduce Sku Complexity - limit one variation of mongodb per versionMark Benvenuto2014-10-163-15/+57
|
* SERVER-14266: Simplify MSI Installation OptionsMark Benvenuto2014-10-165-20/+70
|
* SERVER-14352 rocksdb: Read our writesIgor Canadi2014-10-161-2/+3
| | | | Signed-off-by: Matt Kangas <matt.kangas@mongodb.com>
* SERVER-15375 lazily initShardVersion only when neededGreg Studer2014-10-166-145/+182
|
* SERVER-14425 fixed compilation error in fast_map_noalloc_testBenety Goh2014-10-151-1/+1
|
* SERVER-13635 engine-specific component in collection optionsBenety Goh2014-10-153-0/+105
|
* SERVER-14425 Move all Lock Manager enumerations to a separate headerKaloian Manassiev2014-10-158-104/+97
| | | | | | | Most of the lock manager enumeration types are necessary in places where the entire lock manager interface and its structures are not needed. Also causes circular include dependency with LockStats.
* SERVER-14425 Avoid memory allocations in LockerImplKaloian Manassiev2014-10-153-99/+75
| | | | | | Memory allocations on the common locking path in LockerImpl increase the cost of uncontended lock operations and are fairly easy to be avoided through keeping a stack of already allocated memory.
* SERVER-14425 Implement map with no allocation for the lock manager tracking ↵Kaloian Manassiev2014-10-153-7/+441
| | | | structures
View Lorry Controller Status