Re: Akonadi dsaster
René J.V. Bertin <[email protected]>
| Newsgroups | gmane.comp.kde.users.pim |
|---|---|
| Message-ID | <2329405.p7FNJhUaj5@bola> |
On Monday October 10 2016 10:12:42 Michael Mol wrote: >Here's the relevant excerpt: > >Altogether, here are the changes I've made: > >In my.cnf > >innodb_buffer_pool_size = 1024M # I have 16GB of RAM. I can spare a gig to keep >more of Akonadi's working set in memory. [etc] That's very interesting, but most people will have less RAM in their system (or else need it for other things than "just email"). Do you have recommendations for optimising the mysql settings to a more lowly configuration (say 4GB RAM)? I'd also be interested to know if you have suggestions for when the filesystem is ZFS? On my (ZFS-based) system I have "SET GLOBAL SQL_LOG_BIN=0;" in a dedicated init_file specified in akonadi's mysql.conf, and innodb_use_native_aio=0 . Also: innodb_buffer_pool_size=256M innodb_log_buffer_size=10M innodb_log_file_size=64M query_cache_size=64M plus any tweaks I didn't label in the file. Curiously I have query_cache_type=0; I should experiment with switching it on. With that I'm actually not too unhappy with Akonadi performance, esp. since I did tweak the ZFS dataset on which the db lives a bit, and I have upgraded my hardware (which came with an SSHD; that may have helped). Or maybe I've just grown accustomed to the fact that a full sync takes quite a while... R.