File & Directory Locations¶
Cyrus IMAP maintains several directories for configuration, state and data storage. The locations of these directories and, in many cases, of individual files, may be controlled by settings in imapd.conf(5).
Partitioning is a means to divide storage resources into separate pools and may be defined for the following types of data:
Please consult the documents linked above for more information on these.
You may also find more examples and possible scenarios in the document on Mailbox Distribution.
In general, Cyrus allows one to maintain several separate partitions for each of these data types, and to establish rules governing distribution of data within each pool.
Types of Partitions¶
Cyrus supports several different types of partitions:
Mail Spool Partitions
Each of these are discussed in their own sections of the documentation. All share in common a few aspects in how they are configured. For each partition defined, you must tell Cyrus where the partition is rooted in the filesystem. This is accomplished via a “partition-name:” directive for each partition, where “partition” specifies the partition type and “name” is the actual name of the partition.
Here are some sample declarations of each different type of partition supported within Cyrus. For the purposes of this example, we’ll stipulate the following:
There are two main storage pools, “fast” and “slow,” which are on SSDs and traditional disks, respectively.
The mailboxes are divided alphabetically, with A to M in one group and N to Z in the other.
# The Mail Spool Partitions partition-atom: /var/spool/cyrus/fast/mail/atom/ partition-ntoz: /var/spool/cyrus/fast/mail/ntoz/ # The Metadata Partitions metapartition-atom: /var/spool/cyrus/fast/meta/atom/ metapartition-ntoz: /var/spool/cyrus/fast/meta/ntoz/ # Archive Partitions archivepartition-atom: /var/spool/cyrus/slow/mail/atom/ archivepartition-ntoz: /var/spool/cyrus/slow/mail/ntoz/ # Search Partitions defaultsearchtier: tier1 tier1searchpartition-atom: /var/spool/cyrus/slow/search/atom/ tier1searchpartition-ntoz: /var/spool/cyrus/slow/search/ntoz/
Working With Partitions¶
All partition operations are controlled via settings in imapd.conf(5), and thus require server restarts to effect. There are no administrative commands to manipulate partitions.
Here are the settings for each class of partition:
Mail Spool Partitions¶
While Cyrus itself has no default values for these settings, most
The partition name used by default for new mailboxes. If not specified, the partition with the most free space will be used for new mailboxes.
Note that the partition specified by this option must also be specified as partition-name, where you substitute ‘name’ for the alphanumeric string you set defaultpartition to.
The pathname of the partition name. At least one partition pathname MUST be specified. If the defaultpartition option is used, then its pathname MUST be specified. For example, if the value of the defaultpartion option is part1, then the partition-part1 field is required.
Metadata is information used to process the mailbox data, rather than the contents of the messages withing the mailbox. Examples include headers, caches, indexes, etc.
The pathname of the metadata partition name, corresponding to spool partition partition-name. For any mailbox residing in a directory on partition-name, the metadata files listed in metapartition_files will be stored in a corresponding directory on metapartition-name. Note that not every partition-name option is required to have a corresponding metapartition-name option, so that you can selectively choose which spool partitions will have separate metadata partitions.
Space-separated list of metadata files to be stored on a metapartition rather than in the mailbox directory on a spool partition. Allowed values: header, index, cache, expunge, squat, annotations, lock, dav, archivecache
Cyrus Archiving will migrate mailbox data from the normal mail spool partitions to separate archive partitions, based upon criteria such as age and size. Typical use cases would be to keep so-called “hot” data, such as recent messages, on fast drives, such as SSDs, and migrate “cold” data, such as older or large messages, to slower but cheaper media.
The pathname of the archive partition name, corresponding to spool partition partition-name. For any mailbox residing in a directory on partition-name, the archived messages will be stored in a corresponding directory on archivepartition-name. Note that not every partition-name option is strictly required to have a corresponding archivepartition-name option, but that without one there’s no benefit to enabling archiving.
Modern Cyrus uses the Xapian search engine to index messages for server-side search support. Index data are stored in search “tiers” which are themselves related to search partitions. There are two key settings for search tiers:
Name of the default tier that messages will be indexed to. Search indexes can be organized in tiers to allow index storage in different directories and physical media. See the man page of squatter for details. The default search tier also requires the definition of an according searchtierpartition-name entry.
This option MUST be specified for xapian search.
The pathname where to store the xapian search indexes of searchtier for mailboxes of partition name. This must be configured for the defaultsearchtier and any additional search tier (see squatter for details).
For example: if defaultpartition is defined as part1 and defaultsearchtier as tier1 then the configuration must contain an entry tier1searchpartition-part1 that defines the path where to store this tier1’s search index for the part1 partition.
This option MUST be specified for xapian search.
Back to Administrator Guide