-
Notifications
You must be signed in to change notification settings - Fork 6.4k
EventListener
EventListener class contains a set of call-back functions that will be called when specific RocksDB event happens such as flush. It can be used as a building block for developing custom features such as stats-collector or external compaction algorithm. Available EventListener callbacks can be found in include/rocksdb/listener.h.
In ColumnFamilyOptions, there's a variable called listeners
, which allows developers to add custom EventListener to listen to the events of a specific rocksdb instance or a column family of a rocksdb instance.
// A vector of EventListeners which call-back functions will be called
// when specific RocksDB event happens.
std::vector<std::shared_ptr<EventListener>> listeners;
To listening to a rocksdb instance or a column family of a rocksdb instance, it can be done by simply adding a custom EventListener to ColumnFamilyOptions::listeners
and use that options to open a DB:
// listen to a column family of a rocksdb instance
ColumnFamilyOptions cf_options;
...
cf_options.listeners.emplace_back(new MyListener());
To listen to multiple column families of a rocksdb instance, one can either use a separate Listener instance for each column family,
// one listener for each column family
for (size_t i = 0; i < cf_options.size(); ++i) {
cf_options[i].listeners.emplace_back(new MyListener());
}
or use same Listener instance for all column families:
// one same listener for all column families.
EventListener my_listener = new MyListener();
for (size_t i = 0; i < cf_options.size(); ++i) {
cf_options[i].listeners.emplace_back(my_listener);
}
Note that in either case, unless specially specified in the documentation, all EventListener call-backs must be implemented in a thread-safe way even when an EventListener only listens to a single column family (For example, imagine the case where OnCompactionCompeted()
could be called by multiple threads at the same time as a single column family might complete more than one compaction jobs at the same time.
All EventListener callback will be called using the actual thread that involves in that specific event. For example, it is the RocksDB background flush thread that does the actual flush to call EventListener::OnFlushCompleted()
. This allows developers to collect thread-dependent stats from the EventListener callback such as via thread local variable.
All EventListener callbacks are designed to be called without the current thread holding any DB mutex. This is to prevent potential deadlock and performance issue when using EventListener callback in a complex way. However, that call-back functions should not run for an extended period of time before the function returns, otherwise RocksDB may be blocked. For example, it is not suggested to do DB::CompactFiles()
(as it may run for a long while) or issue many of DB::Put()
(as Put may be blocked in certain cases) in the same thread in the EventListener callback. However, doing DB::CompactFiles()
and DB::Put()
in another thread is considered safe.
Contents
- RocksDB Wiki
- Overview
- RocksDB FAQ
- Terminology
- Requirements
- Contributors' Guide
- Release Methodology
- RocksDB Users and Use Cases
- RocksDB Public Communication and Information Channels
-
Basic Operations
- Iterator
- Prefix seek
- SeekForPrev
- Tailing Iterator
- Compaction Filter
- Multi Column Family Iterator
- Read-Modify-Write (Merge) Operator
- Column Families
- Creating and Ingesting SST files
- Single Delete
- Low Priority Write
- Time to Live (TTL) Support
- Transactions
- Snapshot
- DeleteRange
- Atomic flush
- Read-only and Secondary instances
- Approximate Size
- User-defined Timestamp
- Wide Columns
- BlobDB
- Online Verification
- Options
- MemTable
- Journal
- Cache
- Write Buffer Manager
- Compaction
- SST File Formats
- IO
- Compression
- Full File Checksum and Checksum Handoff
- Background Error Handling
- Huge Page TLB Support
- Tiered Storage (Experimental)
- Logging and Monitoring
- Known Issues
- Troubleshooting Guide
- Tests
- Tools / Utilities
-
Implementation Details
- Delete Stale Files
- Partitioned Index/Filters
- WritePrepared-Transactions
- WriteUnprepared-Transactions
- How we keep track of live SST files
- How we index SST
- Merge Operator Implementation
- RocksDB Repairer
- Write Batch With Index
- Two Phase Commit
- Iterator's Implementation
- Simulation Cache
- [To Be Deprecated] Persistent Read Cache
- DeleteRange Implementation
- unordered_write
- Extending RocksDB
- RocksJava
- Lua
- Performance
- Projects Being Developed
- Misc