When Linden starts, it need a configuration file named "linden.properties" in configuration dir. It specifies all configurable properties in linden.
| Properties | Default Value | Description | Required |
|---|---|---|---|
| port | 9090 | Linden thrift server port | N |
| admin.port | 0 | Linden admin http server port. Set admin.port will enable admin http server and linden http API | N |
| webapp | null | Web app path of linden admin server | Required by admin.port |
| cluster.url | null | Linden zookeeper cluster path | Y |
| shard.id | 0 | Index shard id | N |
| index.directory | null | Path of index directory | Y |
| index.refresh.time | 60 | Seconds between documents is injected to linden and visible to user. | N |
| plugin.path | null | Score model plugin directory path, see score model plugin part at BQL document | N |
| cluster.future.await.timeout | 1000 | Collection timeout of linden instances results in milliseconds | N |
| cluster.future.pool.wait.timeout | 200 | Waiting timeout of linden cluster request future in ExecutorServiceFuturePool | N |
| instance.future.pool.wait.timeout | 200 | Waiting timeout of linden instance request future in ExecutorServiceFuturePool | N |
| enable.parallel.search | true | If enable parallel search | N |
| log.path | null | Linden log directory path | Y |
| enable.cache | false | If enable search cache | N |
| cache.duration | 10 | Cache expiration limit in second | N |
| cache.size | 50000 | Cache size indicate how many recent results are cached | N |
| index.manager.thread.num | 11 | Indexing thread number in index manager | N |
| linden.core.mode | SIMPLE | One linden core corresponds to one physical index directory. There are three linden core modes: SIMPLE: the default model used in linden, only one active linden core and it will never be replaced HOTSWAP: there is always only one active linden core and another linden core is in preparation status. After preparation, prepared linden core can be swapped to active status to replace the previous one MULTI: one shard index is divided to several sub-shard indexes according some rules. Each sub-shard index corresponds to one linden core. See more in Linden Work Mode Document |
N |
| multi.index.division.type | null | The type of multi-core linden mode TIME_HOUR: the index is split in time by hour to sub-index TIME_DAY: the index is split in time by day to sub-index TIME_MONTH: the index is split in time by month to sub-index TIME_YEAR: the index is split in time by year to sub-index DOC_NUM: the index is split by document number, which is defined by multi.index.doc.num.limit INDEX_NAME: the index is split by user customized index name |
N |
| multi.index.doc.num.limit | 10M | Index split document number threshold in multi-core DOC_NUM mode | N |
| multi.index.max.live.index.num | -1 | Linden sub-index number limit of TIME and DOC_NUM division type in multi-index mode, while it doesn’t work for INDEX_NAME division type | N |
| search.time.limit | -1 | Time limit in lucene doc collecting stage, -1 means no limit | N |
| enable.source.field.cache | false | If true prefer fetching source field value in field cache way than stored field from document | N |
| search.thread.pool.json.config | null | Linden cluster and instance search thread pool config, for example: {"cluster":{"min":10,"max":20,"queueSize":1000},"instance":{"min":15,"max":30,"queueSize":2000}} | N |
| gateway.class | null | Gateway plugin, specify the source of index data com.xiaomi.linden.plugin.gateway.kafka.KafkaGateway, which means that linden fetching index data from Kafka. com.xiaomi.linden.plugin.gateway.file.SimpleFileGateway, which means that linden fetching index data from local file. See Linden Plugin Document |
N |
| search.similarity.class | null | Similarity plugin, default similarity is org.apache.lucene.search.similarities.DefaultSimilarity org.apache.lucene.search.similarities.BM25Similarity is another option. See Linde Plugin Document |
N |
| index.analyzer.class | null | Index analyzer plugin, default index analyzer is lucene standard analyzer. See Linden Plugin Document | N |
| search.analyzer.class | null | Search analyzer plugin, default search analyzer is lucene standard analyzer. See Linden Plugin Document | N |
| linden.metric.class | null | Linden metric plugin, which report linden performance data. See Linden Plugin Document | N |
| linden.warmer.class | null | Linden warmer plugin, which will warm linden before linden really serves traffic in restart stage. See Linden Plugin Document | N |
| merge.policy.class | null | Index merge policy plugin. See Linden Plugin Document | N |