Oh, sorry i said index but i mean all files that dovecot create in initial login.
We are trying to make user use horde mail, we are using for webmail here. But the first login took about 5 mins.
I will look up for the most fields in horde login to make the initial cache.
I am using squat, it seens fair enough, is MUCH better or old searchs.
But would you have a better choise ?
Thanks!
Another question, about quota now, some people said dict quota wasn't working very well in dovecot version 1.
What do you thinsk is best quota dict or maildir quota ? We are using maldir for now on.. And the mails are storaged in a shared storage with ocfs2 file system.
Thansk again.
[]'sf.rique
On Mon, Oct 4, 2010 at 5:07 PM, Timo Sirainen tss@iki.fi wrote:
On 4.10.2010, at 20.58, Henrique Fernandes wrote:
First is, is there any command to rebuild index files in dovecot ?
There's a command that is intended to fix index files: doveadm force-rsync. But if your idea was to optimize the initial access it won't help. That's because the main benefits come from the cached fields in dovecot.index.cache file. But it depends on the client what fields it wants cached. You could of course guess some common fields and pre-cache those, like maybe:
doveadm fetch -A 'imap.envelope imap.bodystructure size.virtual date.received' '*' > /dev/null
And i need to force a initial squat build, need comand for it also.
doveadm search -A body asdf
BTW. Have you tried how well Squat works? I'm not happy with its indexing speed.
It is about 5 thousands accounts. Its prefere have this built in dovecot than run a script that log in every sing account to make dovecot index the files, and after it make a small search to make initial squat also.
That's basically what the above commands do, although with less scripting.