zarafa-indexer.cfg

Langue: en

Version: May 2009 (fedora - 01/12/10)

Section: 5 (Format de fichier)

NAME

zarafa-indexer.cfg - The Zarafa Indexer configuration file

SYNOPSIS

indexer.cfg

DESCRIPTION

The indexer.cfg is a configuration file for the Zarafa Indexer Service. indexer.cfg contains instructions for the software to set up the connection, indexing and logging systems.

FILE FORMAT

The file consists of one big section, but parameters can be grouped by functionality.

The parameters are written in the form:

name = value

The file is line-based. Each newline-terminated line represents either a comment, nothing, a parameter or a directive. A line beginning with `#' is considered a comment, and will be ignored by Zarafa. Parameter names are case sensitive. Lines beginning with `!' are directives.

Directives are written in the form:

!directive [argument(s)]

The following directives exist:

include

Include and process argument
Example: !include common.cfg

EXPLANATION OF THE SERVER SETTINGS PARAMETERS

server_socket

Socket to find the connection to the Zarafa server.
Default: file:///var/run/zarafa

run_as_user

After correctly starting, the server process will become this user, dropping root privileges. Note that the log file needs to be writeable by this user, and the directory too to create new logfiles after logrotation. This can also be achieved by setting the correct group and permissions.
Default value is empty, not changing the user after starting.

run_as_group

After correctly starting, the server process will become this group, dropping root privileges.
Default value is empty, not changing the group after starting.

pid_file

Write the process ID number to this file. This is used by the init.d script to correctly stop/restart the service.
Default: /var/run/zarafa-indexer.pid

running_path

Change directory to this path when running in daemonize mode. When using the -F switch to run in the foreground the directory will not be changed.
Default: /

cleanup_lockfiles

If for some unfortunate reason the zarafa-indexer is not cleanly stopped, it may leave lockfiles behind on stores it was processing. To continue the indexing on this possibly corrupt index, you need to remove the lockfile. To restart the index of the store, you need to remove the complete index directory. This directory will be mentioned in the log. You may choose to automatically remove the lockfiles by setting this option to yes. The index will be left on disk.
Default: no

EXPLANATION OF THE SSL SETTINGS PARAMETERS

ssl_key_file

The file containing the private key and certificate. Please read the SSL section in the zarafa-server(1) manual on how to create this file.
Default: /etc/zarafa/ssl/indexer.pem

ssl_key_pass

Enter you password here when your key file contains a password to be readable.
No default set.

ssl_private_key_file

The indexer will use this file as private key for SSL TLS. This file can be created with: openssl genrsa -out /etc/zarafa/indexer/privkey.pem 2048.
Default: /etc/zarafa/indexer/privkey.pem

ssl_certificate_file

The indexer will use this file as certificate for SSL TLS. A self-signed certificate can be created with: openssl req -new -x509 -key /etc/zarafa/indexer/privkey.pem -out /etc/zarafa/indexer/cert.pem -days 1095.
Default: /etc/zarafa/indexer/cert.pem

EXPLANATION OF THE LOGGING SETTINGS PARAMETERS

log_method

The method which should be used for logging. Valid values are:
syslog
Use the Linux system log. All messages will be written to the mail facility. See also syslog.conf(5).

file

Log to a file. The filename will be specified in log_file.

Default: file

log_file

When logging to a file, specify the filename in this parameter. Use - (minus sign) for stderr output.
Default: -

log_level

The level of output for logging in the range from 0 to 5. 0 means no logging, 5 means full logging.
Default: 2

log_timestamp

Specify whether to prefix each log line with a timestamp in 'file' logging mode.
Default: 1

EXPLANATION OF THE INDEXER SETTINGS PARAMETERS

server_bind_name

Connection path to which other processes can connect with the zarafa-indexer for performing search queries.
Use http://0.0.0.0:port to listen as an HTTP service on all network interfaces on the given port number.
Default: file:///var/run/zarafa-indexer

index_path

Base directory under which all index files will be placed, the hierarchy below this folder will be: ${index_path}/${server}/${user}/index
Default: /var/lib/zarafa/index/

index_sync_stream

Enable streaming synchronization. This can increase synchronization speed depending on the setup. During streaming synchronization all messages which will be synchronized will be downloaded in a single call requiring more overall memory. When index_attachments is disabled streaming will cause overhead because all attachments will be downloaded to the client even though they will not be indexed.
This option requires enable_enhanced_ics to be enabled in the zarafa-server.cfg(5), otherwise the indexing will still fallback to the slower synchronization.
Default: yes

index_interval

Interval (in minutes) for indexing. During every interval the zarafa-indexer will synchronize with the zarafa-server to collect all changes for each store and update the index files for the stores accordingly.
Default: 5 minutes

index_threads

Maximum number if indexing threads. In a multi-server environment it is possible to distribute the work for different zarafa-servers over different indexing threads. Note that the work for a single server will be exclusively assigned to a single thread, thus the number of started indexing threads will never exceed the number of zarafa-servers.
Default: 1

index_max_field_length

Maximum number of words from a single message to index. Only the first index_max_field_length from a single message will be indexed, all words above this value will be discarded.
This value is used to control the amount of required memory during the indexing process. More memory will be required during indexing when index_max_field_length is set to a higher value.
Default: 10000

index_merge_factor

Number of index file segments per store before Lucene merges the segments into a single file.
A low value will cause less memory to be used during indexing, but the increased IO access to disk causes the indexing process to be slower, while searching will be faster. A high value will speed up the indexing process while searching will be slower
Batch indexing
If index_interval is set to a high value, set index_merge_factor to a high value (> 10).

Interactive indexing

If index_interval is set to a low value, set index_merge_factor to a low value (< 10).

Default: 10

index_max_buffered_docs

Max number of documents kept in memory before CLucene will write a new index file segment to disk.
Larger values will increase memory usage but makes the indexing process faster.
Default: 10

index_min_merge_docs

Minumum number of messages in a single store which are indexed in memory before the index writer flushes the index to disk as new index file segment.
Creating new index file segments often increases IO access to disk but reduces the amount of memory required during the indexing process.
Default: 10

index_max_merge_docs

Maximum number of documents in a index file segment, when an index file segment contains index_max_merge_docs documents it will no longer be merged with other index file segments.
This will limit the total size of an index file segment but will trigger more index file segments to be created.
Batch indexing
If index_interval is set to a high value, set index_max_merge_docs to a high value (> 10000).

Interactive indexing

If index_interval is set to a low value, set index_max_merge_docs to a low value (< 10000).

Default: 2147483647 (INT32_MAX)

index_term_interval

The fraction of terms in the "dictionary" which should be stored in memory. Smaller values use more memory, but make searching slightly faster, while larger values use less memory and make searching slightly slower. Searching is typically not dominated by dictionary lookup, so tweaking this is rarely useful.
Default: 128

index_cache_timeout

Timeout value (in seconds) for purging all caches used during the indexing process. If set to 0 caching will be disabled.
Default: 0

EXPLANATION OF THE ATTACHMENT INDEXER SETTINGS PARAMETERS

index_attachments

Enable indexing of attachments. When attachments are being indexed, searching for keywords in the body of a message will automatically cause the attachment to be searched as well.
This will slow down the indexing process, require more system memory and increases index file size.
Default: yes

index_attachment_max_size

Maxiumum file size for attachments to be indexed. Any attachment larger then this amount (in kilobytes) will not be indexed.
Default: 5120

index_attachment_parser

For indexing attachment each file must be converted into plain-text by a parser. For this purpose the attachments_parser script has been installed which can be configured to parse specific mime-types.
It is highly advisable to check the attachments_parser.db file to add or remove parser command for specific mime-types and/or file-extensions.
Default: /etc/zarafa/indexerscripts/attachments_parser

index_attachment_parser_max_memory

Limit the maximum amount of memory (in bytes) the parser may use for converting the attachment into plain-text. If this limit is exceeded the parser will be killed and the attachment will not be completely converted. If the value is set to 0 the limit is considered as infinite.
The attachment parser depends on external tools to convert attachments to plain-text, by limiting the amount of resources the parser is allowed to use problems can be prevented when the selected external tool misbehaves with certain attachments.
Default: 0 bytes

index_attachment_parser_max_cputime

Limit the maximum cpu time (in seconds) the parser may use for converting the attachment into plain-text. If this limit is exceeded the parser will be killed and the attachment will not be completely converted. If the value is set to 0 the limit is considered as infinite.
The attachment parser depends on external tools to convert attachments to plain-text, by limiting the amount of resources the parser is allowed to use problems can be prevented when the selected external tool misbehaves with certain attachments.
Default: 0 seconds

EXPLANATION OF THE INDEX FILTERS SETTINGS PARAMETERS

index_block_users

A list of usernames which should not be indexed by the indexing service. Fields in the list are seperated by a SPACE.
Default: empty

index_block_companies

A list of companies which should not be indexed by the indexing service. All users and the public store belonging to these companies will not be indexed. Fields in the list are seperated by a SPACE.
Default: empty

index_allow_servers

A list of server names which should be indexed by the indexing service. All stores located on these servers will be indexed, all other servers will be ignored. If kept empty, no filtering is applied and all detected servers are indexed. Fields in the list are seperated by a SPACE.
Default: empty

RELOADING

The following options are reloadable by sending the zarafa-indexer process a HUP signal:

log_level

index_threads

index_max_field_length

index_merge_factor

index_max_buffered_docs

index_min_merge_docs

index_max_merge_docs

index_term_interval

index_attachments

index_attachment_max_size

index_attachment_parser

index_attachment_parser_max_memory

index_attachment_parser_max_cputime