Re: /var/spool/imap on ZFS
"Niels Dettenbach \(Syndicat IT & Internet\) via Info-cyrus" <[email protected]>
| Newsgroups | gmane.mail.imap.cyrus |
|---|---|
| Message-ID | <[email protected]> |
<[email protected]>: >I'd like to move my current spool from a single (fast) disk to ZFS. Any > >suggestions for the filesystem-parameters? > >Do I want deduplication, for example? Compression? Recordsize? We have a few cyrus-imap on ZFS (BSD) within jails. Each spool got bis own dataset. We do not use compression nor deduplication. On a typically raidz3 setup we use snapshots and replication on ZFS too. But compression could make sense depending from cyrus setup scenario and hardware. Whole spools are on single ZFS dev. I would not spread it "wildly" - means without / instead of cyrus partitioning - over a "combined" ZFS pool from multiple ZFS devs. Use the cyrus scale mech for... Compression could generate strong overheads if you strongly reindex or similiar and most larger attachements are still compressed stuff. Afaik deduplication is not recommended from the ZFS authors in the very most cases - which even plays another roles in ZFS then i.e. in typical RAID. But this are relatively small setups / spools with "just" a few TB spool size where we work with standard record sizes. If they generally grow it may make sense (i would check that) to operate with other parameters - i.e. another record size as well. It would nice to get other experiences as well here... best regards, Niels. -- Niels Dettenbach Syndicat IT & Internet http://www.Syndicat.com ---- Cyrus Home Page: http://www.cyrusimap.org/ List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/ To Unsubscribe: https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus