Manticore Configuration: Searchd

Key DefaultVal Description
access_blob_attrs active mmap, mmap_preread, mlock mmap_preread query, ram Specifies how index's blob attributes file is accessed
access_doclists active file, mmap, mlock file query, ram Specifies how index's doclists file is accessed
access_hitlists active file, mmap, mlock file query, ram Specifies how index's hitlists file is accessed
access_plain_attrs active mmap, mmap_preread, mlock mmap_preread query, ram Specifies how search server will access index's plain attributes
address removed
agent_connect_timeout active int (milliseconds) 1000 (1 second) cluster, net Remote agent connection timeout
agent_query_timeout active int (milliseconds) 3000 (3 seconds) cluster, net Remote agent query timeout
agent_retry_count active int null cluster, net Specifies how many times Manticore will try to connect and query remote agents
agent_retry_delay active int (milliseconds) 500 (1/2 second) cluster, net Specifies the delay before retrying to query a remote agent in case it fails
attr_flush_period active int + timeunit 0 (disable periodic flushing) disk Defines time period between flushing updated attributes to disk
attr_update_reserve active int + sizeunit 128k disk Sets the space to be reserved for blob attribute updates.
auto_optimize active 0, 1, n 1 (enable) optimize Disables or throttles automatic OPTIMIZE.
binlog_flush active 0, 1, 2 2 (flush every transaction, sync every second) disk, log Binary log transaction flush/sync mode
binlog_max_log_size active int bytes 268435456 (256m) disk, log Maximum binary log file size
binlog_path active filepath null (default to build-time directory) log Binary log files path
client_timeout active int + timeunit 5m (5 minutes) net Maximum time to wait between requests when using persistent connections
collation_libc_locale active localename libc_ci charset Server libc locale
collation_server active localename libc_ci charset Default server collation
data_dir active filepath null (must set up schema with config file) base Path to data directory where Manticore stores everything (RT mode)
dist_threads deprecated int 0 (disable in-request parallelism) Max local worker threads to use for parallelizable requests (searching a distributed index; building a batch of snippets)
docstore_cache_size active int + sizeunit 16m query Maximum size of document blocks from document storage that are held in memory
expansion_limit active int 0 (no limit) wildcard, query, nlp Maximum number of expanded keywords for a single wildcard
grouping_in_utc active 0, 1 0 (means 'local tz') timezone, query Turns on using UTC timezone where grouping time fields
ha_period_karma active int (seconds) or + timeunit 60 (1 minute) cluster Agent mirror statistics window size
ha_ping_interval active int (milliseconds) or + timeunit 1000 (1 second) cluster Interval between agent mirror pings
hostname_lookup active request, config_load config_load cluster Hostnames renew strategy
jobs_queue_size active int null (unlimited) query, cluster, net Defines how many "jobs" can be in the queue at the same time
listen active repeatablekey: hostdetails required base Specifies IP address, port or Unix-domain socket path, and modifier for 'vip' status, that searchd will listen on
listen_backlog active int 5 net TCP listen backlog
listen_tfo active 0, 1 1 (enabled) net Allows TCP_FASTOPEN flag for all listeners
log active filepath searchd.log log Path to Manticore server log file
max_batch_queries active int 32 query Limits the amount of queries per batch
max_children removed int 0 (unlimited) Maximum amount of worker threads (or in other words, concurrent queries to run in parallel).
max_connections active int null (unlimited) net Maximum amount of active connections
max_filter_values active int 4096 query Maximum allowed per-filter values count
max_filters active int 256 query Maximum allowed per-query filter count
max_matches removed int 1000 Per-query max matches value.
max_open_files active int null (os limit) sys Maximum num of files which allowed to be opened by server
max_packet_size active int + sizeunit 8m sys Maximum allowed network packet size
max_threads_per_query active int null (searchd.threads) query Instance-wide limit of threads one operation can use. By default appropriate operations can occupy all CPU cores, leaving no room for other operations.
mva_updates_pool removed int + sizeunit 1m Shared pool size for in-memory MVA updates storage
mysql_version_string active string null (return manticore version) base Server version string to return via MySQL protocol
net_send_job active
net_throttle_accept active int 0 (unlimited) sys, net Defines how many clients are accepted on each iteration of the network loop
net_throttle_action active int 0 (unlimited) sys, net Defines how many requests are processed on each iteration of the network loop
net_wait_tm active -1, 0, uint -1 (unlimited) sys, net Controls busy loop interval of a network thread
net_workers active int 1 sys, net Number of network threads
network_timeout active int (seconds) or + timeunit 5 (seconds) sys, net Network timeout for requests from clients
node_address active hostdetails null (same as listen address) cluster Specifies network address of the node
not_terms_only_allowed active 0, 1 query Whether to allow queries with only negation full-text operator.
ondisk_attrs_default removed 0, 1 0 (all attributes are loaded in memory) Instance-wide defaults for ondisk_attrs directive.
ondisk_dict_default removed
optimize_cutoff active int null (# of cpu cores * 2) optimize Sets default index compaction threshold
persistent_connections_limit active int 0 (disable persistent connections) net Maximum number of simultaneous persistent connections to remote persistent agents
pid_file active filepath required base Path to Manticore server pid file
plugin_dir deprecated filepath null (no location) Trusted location for the dynamic libraries (UDFs)
port removed
predicted_time_costs active timecostcalc doc=64, hit=48, skip=2048, match=64 query Costs for the query time prediction model
prefer_rotate deprecated
prefork_rotation_throttle removed
preopen_indexes active 0, 1 1 (preopen all) query Whether to forcibly preopen all indexes on startup
qcache_max_bytes active int (bytes) or + sizeunit 16m query, ram Maximum RAM allocated for cached result sets
qcache_thresh_msec active int 3000 (3 seconds) query, ram Minimum wall time threshold for a query result to be cached
qcache_ttl_sec active int 60 (1 minute) query, ram Expiration period for a cached result set
query_log active filepath null (do not log queries) log Path to query log file
query_log_commands active log
query_log_format active plain, sphinxql plain log Query log format
query_log_min_msec active int 0 (all queries logged) log Prevents logging too fast queries
query_log_mode active permissionslabel 600 log Query log file permissions mode
queue_max_length removed int 0 (unlimited) Maximum pending queries queue length for workers=thread_pool mode
read_buffer deprecated int + sizeunit 256k Per-keyword read buffer size.
read_buffer_columnar active disk, query, columnar
read_buffer_docs active int + sizeunit; min 8k 256k disk, query Per-keyword read buffer size for document lists
read_buffer_hits active int + sizeunit; min 8k 256k disk, query Per-keyword read buffer size for hit lists
read_timeout deprecated int + sizeunit 5 (seconds) Network client request read timeout, in seconds
read_unhinted active int + sizeunit; min 1k 32k disk, query Unhinted read size
rt_flush_period active int (seconds) or + timeunit 10h disk How often Manticore flush real-time indexes' RAM chunks to disk
rt_merge_iops active int 0 (no limit) disk Maximum number of I/O operations (per second) that real-time chunks merging thread is allowed to do
rt_merge_maxiosize active int + sizeunit 0 (no limit) disk Maximum size of an I/O operation that real-time chunks merging thread is allowed to do
seamless_rotate active 0, 1 1 (enable seamless rotation) disk Prevents searchd stalls while rotating indexes with huge amounts of data to precache
server_id active int, 0 - 127 null (randomly-generated) base, cluser Server identifier used as a seed to generate a unique document ID
shutdown_timeout active int, seconds 3 (seconds) sys, base Searchd --stopwait timeout
shutdown_token active sha1 hash null (unused) security, sys, base SHA1 hash of the password required to invoke shutdown command from VIP SQL connection
snippets_file_prefix active filepath null (current working directory) base, query Prefix to prepend to the local file names when generating snippets in load_files mode
sphinxql_state active filepath null (disabled) Path to file where current SQL state will be serialized
sphinxql_timeout active int (seconds) or + timeunit 15m net Maximum time to wait between requests from a mysql client
ssl_ca active filepath null (unused) security Path to SSL Certificate Authority certificate file
ssl_cert active filepath null (unused) security Path to server's SSL certificate
ssl_key active filepath null (unused) security Path to SSL certificate key of the server
subtree_docs_cache active int (k [kilobytes] or m [megabytes], only) 0 (disabled) query Maximum common subtree document cache size
subtree_hits_cache active int (k [kilobytes] or m [megabytes], only) 0 (disabled) query Maximum common subtree hit cache size, per-query
thread_stack active int + sizeunit 128k sys Maximum stack size for a job
threads active int null (number of cpu cores on the server) sys Number of working threads (or, size of thread pool) of Manticore daemon.
unlink_old active 0, 1 1 (unlink) disk Whether to unlink .old index copies on successful rotation
watchdog active 0, 1 1 (enabled) sys, base Whether to enable or disable Manticore server watchdog
workers removed thread_pool, threads thread_pool Multi-processing mode (MPM)
Key DefaultVal Description
access_blob_attrs active mmap, mmap_preread, mlock mmap_preread query, ram Specifies how index's blob attributes file is accessed
access_doclists active file, mmap, mlock file query, ram Specifies how index's doclists file is accessed
access_hitlists active file, mmap, mlock file query, ram Specifies how index's hitlists file is accessed
access_plain_attrs active mmap, mmap_preread, mlock mmap_preread query, ram Specifies how search server will access index's plain attributes
agent active repeatablekey null cluster List of remote agents that are searched every time when the enclosing distributed index is searched.
agent_blackhole active repeatablekey: host:port:indexlist null cluster, net Lets you fire-and-forget queries to remote agents. That is useful for debugging (or just testing) production clusters
agent_connect_timeout active int (milliseconds) 1000 (1 second) cluster, net Remote agent connection timeout
agent_persistent active repeatablekey: host:port:indexlist null cluster, net Asks to persistently connect to agent (i.e. don't drop connection after query).
agent_query_timeout active int (milliseconds) 3000 (3 seconds) cluster, net Remote agent query timeout.
agent_retry_count active int searchd.agent_retry_count cluster, net How many times Manticore will try to connect and query remote agents in distributed index before reporting fatal query error.
attr_update_reserve active int + sizeunit 128k disk Sets the space to be reserved for blob attribute updates.
bigram_freq_words active wordlist null token, nlp, query A list of keywords considered "frequent" when indexing bigrams.
bigram_index active none, all, first_freq, both_freq null token, nlp, query Bigram indexing is a feature to accelerate phrase searches.
blend_chars active charlist null token, nlp Blended characters list.
blend_mode active multiple: trim_none, trim_head, trim_tail, trim_both, trim_all, skip_pure trim_none token. nlp Blended tokens indexing mode.
charset_table active complex value latin and cyrillic characters base, charset, token, nlp Accepted characters array, with case folding rules.
charset_type removed
columnar_attrs active fieldlist specifies what attributes should be stored in the columnar storage instead of the default row-wise storage. columnar, schema
dict active keywords, crc keywords token, nlp The keywords dictionary type.
docinfo removed extern, none, inline extern Docinfo storage mode defines how exactly docinfo will be physically stored on disk and RAM.
docstore_block_size active int + sizeunit 16k disk Size of the block of documents used by document storage.
docstore_compression active lz4, lz4hc, none lz4 disk Type of compression used to compress blocks of documents used by document storage.
docstore_compression_level active 1 - 12 9 disk Compression level in document storage when 'lz4hc' compression is used.
embedded_limit active int + sizeunit 16k parse File size limit for embedded exceptions, wordforms, or stop words file.
enable_star removed
exceptions active filepath null token, nlp, synonym, exception Tokenizing exceptions file.
expand_keywords active 0, 1, exact, star 0 (do not expand keywords) token, nlp, wildcard Expands keywords with their exact forms (i.e. the forms of the keywords before applying any morphological modifications) and/or stars when possible.
global_idf active filepath null cluster, nlp The path to a file with global (cluster-wide) keyword IDFs.
ha_strategy active random, nodeads, noerrors, roundrobin random cluster Agent mirror selection strategy for load balancing.
hitless_words active all, filepath null (all words are considered not hitless) token, nlp Hitless words list.
html_index_attrs active tagattrlist null (do not index markup attributes) html, token A list of markup attributes to index when stripping HTML.
html_remove_elements active taglist null (do not strip contents of any elements) html, token A list of HTML elements for which to strip contents along with the elements themselves.
html_strip active 0, 1 0 (disable stripping) html, token Whether to strip HTML markup from incoming full-text data.
ignore_chars active charlist null token, charset, nlp Ignored characters list.
index_exact_words active 0, 1 0 (do not index) token, stem, wildcard, nlp, morphology Whether to index the original keywords along with the stemmed/remapped versions.
index_field_lengths active 0, 1 0 (do not index) results, query, nlp Enables computing and storing of field lengths (both per-document and average per-index values) into the index.
index_sp active 0, 1 0 (do not detect any sentence or paragraph) token, html Whether to detect and index sentence and paragraph boundaries.
index_token_filter active library name:plugin name:optional string of settings none token, nlp Compiled library for an index-time token filter for index.
index_zones active taglist null (do not index zones) html, token A list of in-field HTML/XML zones to index.
infix_fields active fieldlist null (index all fields in infix mode) token, wildcard, nlp The list of full-text fields to limit infix indexing to.
inplace_docinfo_gap removed int + sizeunit 0 (use separate temporary files). Whether to enable in-place index inversion.
inplace_enable active 0, 1 disk, plain Whether to enable in-place index inversion.
inplace_hit_gap active int + sizeunit disk, plain In-place inversion fine-tuning option. Controls preallocated hitlist gap size.
inplace_reloc_factor active float, between 0 and 1 0.1 disk, plain Controls relocation buffer size within indexing memory arena.
inplace_write_factor active float, between 0 and 1 0.1 disk, plain Controls in-place write buffer size within indexing memory arena.
killlist_target active indexlist + killlist mode (kl / id) null plain Sets the index(es) that the kill-list will be applied to.
local active repeatablekey null cluster, distributed Distributed, remote index reference
max_substring_len deprecated int 0 (do not limit indexed substrings) nlp, wildcard, dictcrc Maximum substring (either prefix or infix) length to index.
min_infix_len active int 0 (disallow infixes) token, wildcard, nlp Minimum infix prefix length to index and search.
min_prefix_len active int 0 (disallow prefixes) token, wildcard, nlp Minimum word prefix length to index and search.
min_stemming_len active int 1 (stem everything) token, wildcard, stem, nlp, morphology Minimum word length at which to enable stemming.
min_word_len active int 1 (index everything) token, nlp Minimum indexed word length.
mirror_retry_count active int 5 cluster The same as agent_retry_count. If both values provided, mirror_retry_count will be taken.
mlock deprecated 0, 1 0 (do not call mlock()) Memory locking for cached data.
morphology active preprocessorlist null (do not apply any preprocessor) token, nlp, morphology A list of morphology preprocessors to apply.
morphology_skip_fields active fieldlist null (apply preprocessors to all fields) token, nlp, morphology A list of fields to skip morphology preprocessing.
ngram_chars active charlist null cjk, nlp N-gram characters list.
ngram_len active 0, 1 0 (disable) cjk, nlp N-gram lengths for N-gram indexing.
ondisk_attrs deprecated 0, 1, pool Allows for fine-grain control over how attributes are loaded into memory when using indexes with external storage.
ondisk_dict removed
overshort_step active 0, 1 1 token, nlp Position increment on overshort (less than min_word_len) keywords.
path active filepath required base, required, cluster Absolute or relative path without extension where to store the index or where to look for it. IMPORTANT: use trailing slash
phrase_boundary active charlist null token, nlp Phrase boundary characters list.
phrase_boundary_step active int token, nlp Phrase boundary word position increment.
prefix_fields active fieldlist null (index all fields in prefix mode) token, wildcard, nlp List of full-text fields to limit prefix indexing to.
preopen active 0, 1 startup, query Tells searchd that it should pre-open all index files on startup (or rotation) and keep them open while it runs.
read_buffer_docs active int + sizeunit; min 8k 256k disk, query Per-keyword read buffer size for document lists.
read_buffer_hits active int + sizeunit; min 8k 256k disk, query Per-keyword read buffer size for hit lists.
read_unhinted active int + sizeunit; min 1k 32k disk, query Unhinted read size.
regexp_filter active repeatablekey: regex null token Regular expressions (regexps) to filter the fields and queries with.
rlp_context removed filepath null RLP context configuration file. Mandatory if RLP is used.
rt_mem_limit active int + sizeunit 128m ram, query RAM chunk size limit.
source active repeatablekey: source null plain Specifies document source to get documents from when the current index is indexed.
stopword_step active 0, 1 1 token, stopword, nlp Position increment on stopwords.
stopwords active filepathlist null token, stopword, nlp Stop word files list (space separated).
stopwords_unstemmed active 0, 1 0 (apply stop word filter after stemming) token, stopword, nlp Whether to apply stop words before or after stemming.
stored_fields active fieldlist null (no fields stored) schema Values from the fields will be both indexed and stored.
stored_only_fields active fieldlist null (no fields stored) schema A list of fields that will be stored in the index but will be not indexed.
type active rt, plain plain base Index type
wordforms active repeatablekey: filepath null token, nlp, wordform Word forms dictionary file.
Key DefaultVal Description
icu_data_dir removed filepath null rlp A folder that contains data used by ICU to segment Chinese text.
json_autoconv_keynames active lowercase null (do not convert anything) json Whether and how to auto-convert key names within JSON attributes
json_autoconv_numbers active 0, 1 0 (do not convert strings into numbers) json Automatically detects and converts possible JSON strings that represent numbers into numeric attributes
lemmatizer_base active filepath /usr/share/manticore nlp Lemmatizer dictionaries base path
on_json_attr_error active fail_index, ignore_attr ignore_attr json, plain What to do if JSON format errors are found
plugin_dir active filepath /usr/local/lib/manticore/ base Location for the dynamic libraries and UDFs
progressive_merge active 0, 1 1 (enabled) disk, optimize Defines order of merging disk chunks in a real-time index
rlp_environment removed filepath null rlp RLP environment configuration file.
rlp_max_batch_docs removed int 50 rlp Maximum number of documents batched before processing them by the RLP.
rlp_max_batch_size removed int 51200 rlp Maximum total size of documents batched before processing them by the RLP.
rlp_root removed filepath null rlp Path to the RLP root folder.
Key DefaultVal Description
ignore_non_plain active 0, 1 0 (don't ignore) plain To ignore warnings about non-plain indexes
json_autoconv_keynames deprecated lowercase null (do not convert anything) json (Moved to Common Section) Whether and how to auto-convert key names within JSON attributes
json_autoconv_numbers deprecated 0, 1 0 (do not convert strings into numbers) json (Moved to Common Section) Automatically detects and converts possible JSON strings that represent numbers into numeric attributes
lemmatizer_cache active int + sizeunit 256k ram Lemmatizer cache size
max_file_field_buffer active int + sizeunit (min 1m) 1m ram, plain Maximum file field adaptive buffer size
max_iops active int 0 (unlimited) disk Maximum indexation I/O operations per second
max_iosize active int 0 (unlimited) disk Maximum allowed I/O operation size
max_xmlpipe2_field active int 0 (unlimited) disk, plain Maximum allowed field size for XMLpipe2 source type
mem_limit active int + sizeunit (min 256k, max 2047m) 128m ram, plain Indexing RAM usage limit
on_file_field_error active ignore_field, skip_document, fail_index ignore_field indexing How to handle IO errors in file fields
on_json_attr_error deprecated fail_index, ignore_attr ignore_attr json, plain (Moved to Common Section) What to do if JSON format errors are found
write_buffer active int + sizeunit 1m ram Write buffer size