Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | checkpoint work for SERVER-4644 | U-tellus\cwestin | 2012-02-14 | 7 | -19/+367 |
| | |||||
* | first cut at named traces (unused); copy ctor and assignment op for BSONObj | U-tellus\cwestin | 2012-02-14 | 3 | -2/+15 |
| | |||||
* | checkpoint dependency tracker for SERVER-4644 | U-tellus\cwestin | 2012-02-14 | 23 | -53/+577 |
| | |||||
* | SERVER-3632 use Windows syntax for default dbpath on Windows | Eric Milkie | 2012-02-14 | 2 | -4/+4 |
| | | | | Note that we still use the current disk for \data\db (not necessarily C:) | ||||
* | SERVER-2601: Get shared client build working on Linux. | Andy Schwerin | 2012-02-14 | 1 | -2/+2 |
| | | | | | "util/version.cpp" should not be part of the client build. Linking of the shared library form of the client failed, because this file was being included. | ||||
* | SERVER-4731 use simpler english | Eric Milkie | 2012-02-14 | 1 | -1/+1 |
| | |||||
* | added inlines for compiling with older gcc | Kevin Matulef | 2012-02-14 | 1 | -4/+4 |
| | |||||
* | adding MurmurHash3 library | Kevin Matulef | 2012-02-14 | 5 | -2/+377 |
| | |||||
* | SERVER-4731 add new state for removed replicaset members | Eric Milkie | 2012-02-14 | 6 | -21/+37 |
| | | | | | | | New behavior: when you remove a replica set node, its state changes to SHUNNED until it is either shut down or readded to a replica set. After it is removed, the removed node sees only itself in the replica set configuration. | ||||
* | Revert "SERVER-4957: If shard lookup fails, retry appending the default port." | Spencer T Brody | 2012-02-14 | 1 | -6/+0 |
| | | | | This reverts commit 444658ee70b45f7678fd60553e300c5f0563d017. | ||||
* | prevent windows from treating max as a macro SERVER-4367 | Mathias Stearn | 2012-02-13 | 1 | -1/+1 |
| | |||||
* | fix Windows build by deactivating max macro | Eric Milkie | 2012-02-13 | 1 | -0/+1 |
| | | | | | The better fix for this would be to include pch.h in rwlockimpl.cpp. But first I need to investigate why it wasn't included originally. | ||||
* | Use Task API correctly in mongos SERVER-4367 | Mathias Stearn | 2012-02-13 | 2 | -4/+3 |
| | |||||
* | typo MINOR | Mathias Stearn | 2012-02-13 | 1 | -1/+1 |
| | |||||
* | Check limits when converting Date_t to time_t SERVER-4834 | Mathias Stearn | 2012-02-13 | 1 | -2/+9 |
| | |||||
* | SERVER-4957: If shard lookup fails, retry appending the default port. | Spencer T Brody | 2012-02-13 | 1 | -0/+6 |
| | |||||
* | Better log message SERVER-4810 | Kristina | 2012-02-13 | 1 | -1/+4 |
| | |||||
* | Check optimes before stepping down due to priority SERVER-4810 | Kristina | 2012-02-13 | 1 | -1/+4 |
| | |||||
* | Extend writelock during stepdown to include socket close SERVER-4619 | Kristina | 2012-02-13 | 1 | -5/+1 |
| | |||||
* | Added array of success/failures to applyOps output SERVER-4259 | Kristina | 2012-02-13 | 1 | -2/+4 |
| | |||||
* | SERVER-4892 - Perform correct authorization checks when doing killCursors. | Andy Schwerin | 2012-02-13 | 3 | -44/+50 |
| | | | | | | | | | | Prior to this patch, we were incorrectly interpreting a portion of the dbKillCursors message as a namespace string, for auth purposes. This was nonsense, as each cursor listed in the killCursors could be from a distinct namespace. In this patch, the client's authority to kill each cursor is checked as part of ClientCursor::erase. Failure to erase due to lack of authorization is reported identically to failure because the cursor never existed or no longer exists. Test to follow. | ||||
* | Remove unused, incorrect and misleading struct OP_GETMORE. | Andy Schwerin | 2012-02-13 | 1 | -7/+0 |
| | |||||
* | SERVER-4212: Added localTime to ismaster command | Jared Rosoff | 2012-02-13 | 1 | -0/+1 |
| | |||||
* | use correct assert in update SERVER-4938 | Eliot Horowitz | 2012-02-12 | 1 | -3/+1 |
| | |||||
* | SERVER-4948 disable jsobj fuzz tests | Aaron | 2012-02-12 | 1 | -0/+3 |
| | |||||
* | don't add tests to suite multiple times | Aaron | 2012-02-12 | 1 | -2/+0 |
| | |||||
* | SERVER-3719 specify correct jira number | Aaron | 2012-02-12 | 1 | -5/+5 |
| | |||||
* | use str::equals | Aaron | 2012-02-12 | 1 | -1/+1 |
| | |||||
* | typo in log message | Aaron | 2012-02-12 | 1 | -1/+1 |
| | |||||
* | fix assertion text | Aaron | 2012-02-12 | 1 | -1/+1 |
| | |||||
* | SERVER-4907 hint _id or else $natural when calling updateObjects for oplog ↵ | Aaron | 2012-02-12 | 12 | -152/+377 |
| | | | | entries | ||||
* | fix argument passing support in unit test framework by holding a value ↵ | Aaron | 2012-02-12 | 1 | -1/+1 |
| | | | | rather than a reference in the test holder | ||||
* | pass hint using BSONObj rather than BSONElement | Aaron | 2012-02-12 | 6 | -56/+58 |
| | |||||
* | query optimizer cursor add check that a recorded unindexed cursor will ↵ | Aaron | 2012-02-12 | 1 | -0/+21 |
| | | | | trigger plan reevaluation in requireIndex mode | ||||
* | Clarify oplog behavior of dbhelpers functions | Aaron | 2012-02-12 | 1 | -4/+10 |
| | |||||
* | remove unused DbSet code | Aaron | 2012-02-12 | 2 | -73/+0 |
| | |||||
* | Use shard name instead of connection string for split and move chunk commands. | Spencer T Brody | 2012-02-10 | 3 | -22/+11 |
| | | | | Also don't look up shards by replica set names. | ||||
* | Avoid asserts in message_port.cpp when bridge gets shut down by unit tests | Eric Milkie | 2012-02-10 | 1 | -0/+4 |
| | |||||
* | SERVER-4333 add fromMigrate flag to identify sharding oplog entries | Eric Milkie | 2012-02-10 | 9 | -39/+90 |
| | |||||
* | Add unit test for SERVER-4745. | Spencer T Brody | 2012-02-09 | 1 | -0/+22 |
| | |||||
* | Remove check for BtrFS SERVER-4885 | Mathias Stearn | 2012-02-09 | 1 | -4/+1 |
| | | | | As of 2.6.31 it supports fallocate so don't need to use sparse files anyway | ||||
* | Add -Woverloaded-virtual to CXXFLAGS and fix build SERVER-4802 | Mathias Stearn | 2012-02-09 | 4 | -5/+5 |
| | |||||
* | Use long long in key parts of OpDebug | Mathias Stearn | 2012-02-09 | 2 | -4/+4 |
| | |||||
* | Add nupdated and nmoved to profile object and slow op logging SERVER-4860 | Mathias Stearn | 2012-02-09 | 4 | -6/+18 |
| | | | | leaving bool moved in profiler object for backwards compatibility | ||||
* | goodbye boost::call_once | Mathias Stearn | 2012-02-09 | 2 | -13/+15 |
| | |||||
* | Don't destroy _hostNameCached SERVER-4367 | Mathias Stearn | 2012-02-09 | 2 | -2/+2 |
| | |||||
* | Better codegen for inline common-case version of spin_lock | Mathias Stearn | 2012-02-09 | 1 | -1/+1 |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The old code would jump over the call to _lk if pthread_spin_lock worked. Now it only jumps if it fails. Old: 0x0000000000a4cc08 <+40>: callq 0x7a06d0 <pthread_spin_trylock@plt> 0x0000000000a4cc0d <+45>: test %eax,%eax 0x0000000000a4cc0f <+47>: je 0xa4cc19 <mongo::getHostNameCached()+57> 0x0000000000a4cc11 <+49>: mov %r12,%rdi 0x0000000000a4cc14 <+52>: callq 0xa49db0 <mongo::SpinLock::_lk()> 0x0000000000a4cc19 <+57>: mov 0x8(%r12),%rax 0x0000000000a4cc1e <+62>: mov %r12,%rdi 0x0000000000a4cc21 <+65>: mov -0x18(%rax),%r13 0x0000000000a4cc25 <+69>: callq 0x7a06e0 <pthread_spin_unlock@plt> New: 0x0000000000ce4468 <+40>: callq 0x7a06c0 <pthread_spin_trylock@plt> 0x0000000000ce446d <+45>: test %eax,%eax 0x0000000000ce446f <+47>: jne 0xce4544 <mongo::getHostNameCached()+260> 0x0000000000ce4475 <+53>: mov 0x8(%r12),%rax 0x0000000000ce447a <+58>: mov %r12,%rdi 0x0000000000ce447d <+61>: mov -0x18(%rax),%r13 0x0000000000ce4481 <+65>: callq 0x7a06d0 <pthread_spin_unlock@plt> ... 0x0000000000ce4544 <+260>: mov %r12,%rdi 0x0000000000ce4547 <+263>: callq 0xce1610 <mongo::SpinLock::_lk()> 0x0000000000ce454c <+268>: nopl 0x0(%rax) 0x0000000000ce4550 <+272>: jmpq 0xce4475 <mongo::getHostNameCached()+53> | ||||
* | SERVER-4806: check for BufBuilder realloc success. (thx @spencerjackson for ↵ | Ben Becker | 2012-02-09 | 1 | -1/+3 |
| | | | | pull req 169) | ||||
* | SERVER-4866 include the webserver in our attempt to prevent socket leaks | Eric Milkie | 2012-02-09 | 2 | -9/+9 |
| | | | | This fixes httpClientTest | ||||
* | SERVER-4866 prevent resource leaks by ensuring close() is always called for ↵ | Eric Milkie | 2012-02-09 | 7 | -30/+44 |
| | | | | Socket |