with tuning, we can drop this: according to upstream: [mariadb] key_buffer_size = 1m # (as it's not used; no aria and no myisam) aria_pagecache_buffer_size = 4m (as aria is used for tmp tables) aria_sort_buffer_size = 4m (same as above) it's in a separate section, so that mysql would have no issue with it, if we were later to use mysql...
CC: (none) => alien
correction: [mariadb] key_buffer_size = 128k # (as it's not used; no aria and no myisam) aria_pagecache_buffer_size = 4m (as aria is used for tmp tables) aria_sort_buffer_size = 4m (same as above) (so this can be appended to the mysql.conf that gets "generated" for .local
Hi, thanks for reporting this bug. Should this bug not be reporting upstream ?
Keywords: (none) => TriagedAssignee: bugsquad => balcaen.john
i have no idea, i was maybe planning on maybe doing this myself, if mikala agreed...
ok, after more discussions it didn't seem to have much effect (3MB), so we should go for: [mariadb] key_buffer_size = 8m aria_pagecache_buffer_size = 8m aria_sort_buffer_size = 8m
Ok without thoses lines mysqld is using 146 692 k & 7012 of shared memory against 124 576 & 7240 of shared memory when using specific maria db. However it seems that mysqld is now using a little more CPU than before thoses configuration settings.
CC: (none) => balcaen.john
you think we should leave it at default settings? or do you think your case is so advanced that we should still keep these and let the "advanced" users tune it? it's your call... if you want to get this tuned to perfection, i'm pretty sure upstream will be quite accomodating... (there are some people there who will go to the kB of memory usage)
Yep we should leave it as default upstream & eventually report it upstream (akonadi) to have their opinion. Closing as wontfix.
Status: NEW => RESOLVEDResolution: (none) => WONTFIX