davidradl commented on code in PR #26107: URL: https://github.com/apache/flink/pull/26107#discussion_r1944989494
########## docs/layouts/shortcodes/generated/forst_configurable_configuration.html: ########## @@ -0,0 +1,168 @@ +<table class="configuration table table-bordered"> + <thead> + <tr> + <th class="text-left" style="width: 20%">Key</th> + <th class="text-left" style="width: 15%">Default</th> + <th class="text-left" style="width: 10%">Type</th> + <th class="text-left" style="width: 55%">Description</th> + </tr> + </thead> + <tbody> + <tr> + <td><h5>state.backend.forst.block.blocksize</h5></td> + <td style="word-wrap: break-word;">4 kb</td> + <td>MemorySize</td> + <td>The approximate size (in bytes) of user data packed per block. The default blocksize is '4KB'.</td> + </tr> + <tr> + <td><h5>state.backend.forst.block.cache-size</h5></td> + <td style="word-wrap: break-word;">8 mb</td> + <td>MemorySize</td> + <td>The amount of the cache for data blocks in ForSt. The default block-cache size is '8MB'.</td> + </tr> + <tr> + <td><h5>state.backend.forst.block.metadata-blocksize</h5></td> + <td style="word-wrap: break-word;">4 kb</td> + <td>MemorySize</td> + <td>Approximate size of partitioned metadata packed per block. Currently applied to indexes block when partitioned index/filters option is enabled. The default blocksize is '4KB'.</td> + </tr> + <tr> + <td><h5>state.backend.forst.bloom-filter.bits-per-key</h5></td> + <td style="word-wrap: break-word;">10.0</td> + <td>Double</td> + <td>Bits per key that bloom filter will use, this only take effect when bloom filter is used. The default value is 10.0.</td> + </tr> + <tr> + <td><h5>state.backend.forst.bloom-filter.block-based-mode</h5></td> + <td style="word-wrap: break-word;">false</td> + <td>Boolean</td> + <td>If true, ForSt will use block-based filter instead of full filter, this only take effect when bloom filter is used. The default value is 'false'.</td> + </tr> + <tr> + <td><h5>state.backend.forst.compaction.filter.periodic-compaction-time</h5></td> + <td style="word-wrap: break-word;">30 d</td> + <td>Duration</td> + <td>Periodic compaction could speed up expired state entries cleanup, especially for state entries rarely accessed. Files older than this value will be picked up for compaction, and re-written to the same level as they were before. It makes sure a file goes through compaction filters periodically. 0 means turning off periodic compaction.The default value is '30days'.</td> + </tr> + <tr> + <td><h5>state.backend.forst.compaction.filter.query-time-after-num-entries</h5></td> + <td style="word-wrap: break-word;">1000</td> + <td>Long</td> + <td>Number of state entries to process by compaction filter before updating current timestamp. Updating the timestamp more often can improve cleanup speed, but it decreases compaction performance because it uses JNI calls from native code.The default value is '1000L'.</td> + </tr> + <tr> + <td><h5>state.backend.forst.compaction.level.max-size-level-base</h5></td> + <td style="word-wrap: break-word;">256 mb</td> + <td>MemorySize</td> + <td>The upper-bound of the total size of level base files in bytes. The default value is '256MB'.</td> + </tr> + <tr> + <td><h5>state.backend.forst.compaction.level.target-file-size-base</h5></td> + <td style="word-wrap: break-word;">64 mb</td> + <td>MemorySize</td> + <td>The target file size for compaction, which determines a level-1 file size. The default value is '64MB'.</td> + </tr> + <tr> + <td><h5>state.backend.forst.compaction.level.use-dynamic-size</h5></td> + <td style="word-wrap: break-word;">false</td> + <td>Boolean</td> + <td>If true, ForSt will pick target size of each level dynamically. From an empty DB, ForSt would make last level the base level, which means merging L0 data into the last level, until it exceeds max_bytes_for_level_base. And then repeat this process for second last level and so on. The default value is 'false'. For more information, please refer to <a href="https://github.com/facebook/rocksdb/wiki/Leveled-Compaction#level_compaction_dynamic_level_bytes-is-true";>RocksDB's doc.</a></td> + </tr> + <tr> + <td><h5>state.backend.forst.compaction.style</h5></td> + <td style="word-wrap: break-word;">LEVEL</td> + <td><p>Enum</p></td> + <td>The specified compaction style for DB. Candidate compaction style is LEVEL, FIFO, UNIVERSAL or NONE, and Flink chooses 'LEVEL' as default style.<br /><br />Possible values:<ul><li>"LEVEL"</li><li>"UNIVERSAL"</li><li>"FIFO"</li><li>"NONE"</li></ul></td> + </tr> + <tr> + <td><h5>state.backend.forst.compression.per.level</h5></td> + <td style="word-wrap: break-word;">SNAPPY_COMPRESSION</td> + <td><p>List<Enum></p></td> + <td>A semicolon-separated list of Compression Type. Different levels can have different compression policies. In many cases, lower levels use fast compression algorithms, while higher levels with more data use slower but more effective compression algorithms. The N th element in the List corresponds to the compression type of the level N-1When <code class="highlighter-rouge">state.backend.forst.compaction.level.use-dynamic-size</code> is true, compression_per_level[0] still determines L0, but other elements are based on the base level and may not match the level seen in the info log<br />Note: If the List size is smaller than the level number, the undefined lower level uses the last Compression Type in the List<br />Some commonly used compression algorithms for candidates include <code class="highlighter-rouge">NO_COMPRESSION</code> ,<code class="highlighter-rouge">SNAPPY_COMPRESSION</code> and <code class="highlighter-rouge">LZ4_COMPRESSION</code><br />The default value is <code class="highlighter-rouge">SNAPPY_COMPRESSION</code>, which means that all data uses the Snappy compression algorithm.Likewise, if set to <code class="highlighter-rouge">NO_COMPRESSION</code> , means that all data is not compressed, which will achieve faster speed but will bring some space amplification.In addition, if we need to consider both spatial amplification and performance, we can also set it to '<code class="highlighter-rouge">NO_COMPRESSION</code>;<code class="highlighter-rouge">NO_COMPRESSION</code>;<code class="highlighter-rouge">LZ4_COMPRESSION</code>', which means that L0 and L1 data will not be compressed, and other data will be compressed using LZ4.<br /><br />Possible values:<ul><li>"NO_COMPRESSION"</li><li>"SNAPPY_COMPRESSION"</li><li>"ZLIB_COMPRESSION"</li><li>"BZLIB2_COMPRESSION"</li><li>"LZ4_COMPRESSION"</li><li>"LZ4HC_COMPRESSION"</li><li>"XPRESS_COMPRESSION"</li><li>"ZSTD_COMPRESSION"</li><li>"DISABLE_COMPRESSION_OPTION"</li></ul></td> + </tr> + <tr> + <td><h5>state.backend.forst.files.open</h5></td> + <td style="word-wrap: break-word;">-1</td> + <td>Integer</td> + <td>The maximum number of open files (per stateful operator) that can be used by the DB, '-1' means no limit. The default value is '-1'.</td> + </tr> + <tr> + <td><h5>state.backend.forst.log.dir</h5></td> + <td style="word-wrap: break-word;">(none)</td> + <td>String</td> + <td>The directory for ForSt's information logging files. If empty (Flink default setting), log files will be in the same directory as the Flink log. If non-empty, this directory will be used and the data directory's absolute path will be used as the prefix of the log file name. If setting this option as a non-existing location, e.g '/dev/null', ForSt will then create the log under its own database folder as before.</td> + </tr> + <tr> + <td><h5>state.backend.forst.log.file-num</h5></td> + <td style="word-wrap: break-word;">4</td> + <td>Integer</td> + <td>The maximum number of files ForSt should keep for information logging (Default setting: 4).</td> + </tr> + <tr> + <td><h5>state.backend.forst.log.level</h5></td> + <td style="word-wrap: break-word;">INFO_LEVEL</td> + <td><p>Enum</p></td> + <td>The specified information logging level for ForSt. If unset, Flink will use <code class="highlighter-rouge">INFO_LEVEL</code>.<br />Note: ForSt info logs will not be written to the TaskManager logs and there is no rolling strategy, unless you configure <code class="highlighter-rouge">state.backend.forst.log.dir</code>, <code class="highlighter-rouge">state.backend.forst.log.max-file-size</code>, and <code class="highlighter-rouge">state.backend.forst.log.file-num</code> accordingly. Without a rolling strategy, long-running tasks may lead to uncontrolled disk space usage if configured with increased log levels!<br />There is no need to modify the ForSt log level, unless for troubleshooting ForSt.<br /><br />Possible values:<ul><li>"DEBUG_LEVEL"</li><li>"INFO_LEVEL"</li><li>"WARN_LEVEL"</li><li>"ERROR_LEVEL"</li><li>"FATAL_LEVEL"</li><li>"HEADER_LEVEL"</li><li>"NUM_INFO_LOG_LEVELS"</li></ul></td> + </tr> + <tr> + <td><h5>state.backend.forst.log.max-file-size</h5></td> + <td style="word-wrap: break-word;">25 mb</td> + <td>MemorySize</td> + <td>The maximum size of ForSt's file used for information logging. If the log files becomes larger than this, a new file will be created. If 0, all logs will be written to one log file. The default maximum file size is '25MB'. </td> + </tr> + <tr> + <td><h5>state.backend.forst.rescaling.use-delete-files-in-range</h5></td> + <td style="word-wrap: break-word;">false</td> + <td>Boolean</td> + <td>If true, during rescaling, the deleteFilesInRange API will be invoked to clean up the useless files so that local disk space can be reclaimed more promptly.</td> + </tr> + <tr> + <td><h5>state.backend.forst.restore-overlap-fraction-threshold</h5></td> + <td style="word-wrap: break-word;">0.0</td> + <td>Double</td> + <td>The threshold of overlap fraction between the handle's key-group range and target key-group range. When restore base DB, only the handle which overlap fraction greater than or equal to threshold has a chance to be an initial handle. The default value is 0.0, there is always a handle will be selected for initialization. </td> + </tr> + <tr> + <td><h5>state.backend.forst.thread.num</h5></td> + <td style="word-wrap: break-word;">2</td> + <td>Integer</td> + <td>The maximum number of concurrent background flush and compaction jobs (per stateful operator). The default value is '2'.</td> + </tr> + <tr> + <td><h5>state.backend.forst.use-bloom-filter</h5></td> + <td style="word-wrap: break-word;">false</td> + <td>Boolean</td> + <td>If true, every newly created SST file will contain a Bloom filter. It is disabled by default.</td> + </tr> + <tr> + <td><h5>state.backend.forst.use-ingest-db-restore-mode</h5></td> + <td style="word-wrap: break-word;">false</td> + <td>Boolean</td> + <td>A recovery mode that directly clips and ingests multiple DBs during state recovery if the keys in the SST files does not exceed the declared key-group range.</td> Review Comment: I assume the test refers to whetehr this is true. What about when it is false. Also guidance on this would be good. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
