发布于 2015-09-14 14:54:28 | 154 次阅读 | 评论: 0 | 来源: 网络整理
MongoDB uses write ahead logging to an on-disk journal to guarantee write operation durability and to provide crash resiliency. Before applying a change to the data files, MongoDB writes the change operation to the journal. If MongoDB should terminate or encounter an error before it can write the changes from the journal to the data files, MongoDB can re-apply the write operation and maintain a consistent state.
Without a journal, if mongod exits unexpectedly, you must assume your data is in an inconsistent state, and you must run either repair or, preferably, resync from a clean member of the replica set.
With journaling enabled, if mongod stops unexpectedly, the program can recover everything written to the journal, and the data remains in a consistent state. By default, the greatest extent of lost writes, i.e., those not made to the journal, is no more than the last 100 milliseconds.
With journaling, if you want a data set to reside entirely in RAM, you need enough RAM to hold the dataset plus the “write working set.” The “write working set” is the amount of unique data you expect to see written between re-mappings of the private view. For information on views, see Storage Views used in Journaling.
重要
在 2.0 版更改: For 64-bit builds of mongod, journaling is enabled by default. For other platforms, see journal.
在 2.0 版更改: For 64-bit builds of mongod, journaling is enabled by default.
To enable journaling, start mongod with the --journal command line option.
If no journal files exist, when mongod starts, it must preallocate new journal files. During this operation, the mongod is not listening for connections until preallocation completes: for some systems this may take a several minutes. During this period your applications and the mongo shell are not available.
警告
Do not disable journaling on production systems. If your mongod instance stops without shutting down cleanly unexpectedly for any reason, (e.g. power failure) and you are not running with journaling, then you must recover from an unaffected replica set member or backup, as described in repair.
To disable journaling, start mongod with the --nojournal command line option.
You can get commit acknowledgment with the getLastError command and the j option. For details, see Internal Operation of Write Concern.
To avoid preallocation lag, you can preallocate files in the journal directory by copying them from another instance of mongod.
Preallocated files do not contain data. It is safe to later remove them. But if you restart mongod with journaling, mongod will create them again.
Example
The following sequence preallocates journal files for an instance of mongod running on port 27017 with a database path of /data/db.
For demonstration purposes, the sequence starts by creating a set of journal files in the usual way.
Create a temporary directory into which to create a set of journal files:
mkdir ~/tmpDbpath
Create a set of journal files by staring a mongod instance that uses the temporary directory:
mongod --port 10000 --dbpath ~/tmpDbpath --journal
When you see the following log output, indicating mongod has the files, press CONTROL+C to stop the mongod instance:
web admin interface listening on port 11000
Preallocate journal files for the new instance of mongod by moving the journal files from the data directory of the existing instance to the data directory of the new instance:
mv ~/tmpDbpath/journal /data/db/
Start the new mongod instance:
mongod --port 27017 --dbpath /data/db --journal
Use the following commands and methods to monitor journal status:
The serverStatus command returns database status information that is useful for assessing performance.
Use journalLatencyTest to measure how long it takes on your volume to write to the disk in an append-only fashion. You can run this command on an idle system to get a baseline sync time for journaling. You can also run this command on a busy system to see the sync time on a busy system, which may be higher if the journal directory is on the same volume as the data files.
The journalLatencyTest command also provides a way to check if your disk drive is buffering writes in its local cache. If the number is very low (i.e., less than 2 milliseconds) and the drive is non-SSD, the drive is probably buffering writes. In that case, enable cache write-through for the device in your operating system, unless you have a disk controller card with battery backed RAM.
在 2.0 版更改.
You can set the group commit interval using the --journalCommitInterval command line option. The allowed range is 2 to 300 milliseconds.
Lower values increase the durability of the journal at the expense of disk performance.
On a restart after a crash, MongoDB replays all journal files in the journal directory before the server becomes available. If MongoDB must replay journal files, mongod notes these events in the log output.
There is no reason to run repairDatabase in these situations.
When running with journaling, MongoDB stores and applies write operations in memory and in the journal before the changes are in the data files.
With journaling enabled, MongoDB creates a journal directory within the directory defined by dbpath, which is /data/db by default. The journal directory holds journal files, which contain write-ahead redo logs. The directory also holds a last-sequence-number file. A clean shutdown removes all the files in the journal directory.
Journal files are append-only files and have file names prefixed with j._. When a journal file holds 1 gigabyte of data, MongoDB creates a new journal file. Once MongoDB applies all the write operations in the journal files, it deletes these files. Unless you write many bytes of data per-second, the journal directory should contain only two or three journal files.
To limit the size of each journal file to 128 megabytes, use the smallfiles run time option when starting mongod.
To speed the frequent sequential writes that occur to the current journal file, you can ensure that the journal directory is on a different system.
重要
If you place the journal on a different filesystem from your data files you cannot use a filesystem snapshot to capture consistent backups of a dbpath directory.
注解
Depending on your file system, you might experience a preallocation lag the first time you start a mongod instance with journaling enabled. MongoDB preallocates journal files if it is faster on your file system to create files of a pre-defined. The amount of time required to pre-allocate lag might last several minutes, during which you will not be able to connect to the database. This is a one-time preallocation and does not occur with future invocations.
To avoid preallocation lag, see Avoid Preallocation Lag.
Journaling adds three storage views to MongoDB.
The shared view stores modified data for upload to the MongoDB data files. The shared view is the only view with direct access to the MongoDB data files. When running with journaling, mongod asks the operating system to map your existing on-disk data files to the shared view memory view. The operating system maps the files but does not load them. MongoDB later loads data files to shared view as needed.
The private view stores data for use in read operations. MongoDB maps private view to the shared view and is the first place MongoDB applies new write operations.
The journal is an on-disk view that stores new write operations after MongoDB applies the operation to the private cache but before applying them to the data files. The journal provides durability. If the mongod instance were to crash without having applied the writes to the data files, the journal could replay the writes to the shared view for eventual upload to the data files.
MongoDB copies the write operations to the journal in batches called group commits. By default, MongoDB performs a group commit every 100 milliseconds: as a result MongoDB commits all operations within a 100 millisecond window in a single batch. These “group commits” help minimize the performance impact of journaling.
Journaling stores raw operations that allow MongoDB to reconstruct the following:
As write operations occur, MongoDB writes the data to the private view in RAM and then copies the write operations in batches to the journal. The journal stores the operations on disk to ensure durability. MongoDB adds the operations as entries on the journal’s forward pointer. Each entry describes which bytes the write operation changed in the data files.
MongoDB next applies the journal’s write operations to the shared view. At this point, the shared view becomes inconsistent with the data files.
At default intervals of 60 seconds, MongoDB asks the operating system to flush the shared view to disk. This brings the data files up-to-date with the latest write operations.
When MongoDB flushes write operations to the data files, MongoDB removes the write operations from the journal’s behind pointer. The behind pointer is always far back from advanced pointer.
As part of journaling, MongoDB routinely asks the operating system to remap the shared view to the private view, for consistency.
注解
The interaction between the shared view and the on-disk data files is similar to how MongoDB works without journaling, which is that MongoDB asks the operating system to flush in-memory changes back to the data files every 60 seconds.