Re: Search Folder size - Xapian enabled

"Robert Stepanek" <[email protected]>
Newsgroups gmane.mail.imap.cyrus
Message-ID <[email protected]>
On Tue, Mar 22, 2022, at 4:05 PM, Tecnologia UNNOBA wrote:
> I can't find "xapian-version" that you mention.
> 
> Instead:
> 
> server1:/var/spool/cyrus/mail# xapian-check --version
> xapian-check - xapian-core 1.4.18
> 

Yes, sorry. That's the binary I meant.

> I didn't compacted those databases. I founded some perl scripts on internet that execute via crontab, but I didn't know if cyrus can/must do it himself.
> 

Cyrus can do that with the squatter executable. Before doing that it is better to though to look at your database as they are. The xapian-delve binary which comes with your Xapian installation is the tool for that.

First, inspect the output of the high-level metadata, like

# xapian-delve 1626300101/search/user/uuid/9/9/99eee79e-3317-49cb-87ac-4db68335e450/xapian/
UUID = bee410b2-1d68-485c-9e8e-86747d7f4b82
number of documents = 2
average document length = 11
document length lower bound = 2
document length upper bound = 20
highest document id ever used = 2
has positional information = true
revision = 2
currently open for writing = false 

Does the number of documents look reasonable compared with the number of emails? It's ok for them to be twice as much, e.g. if your emails typically have a plain text and HTML part. In case you are not indexing attachments (which I guess you do not), then a much larger number of documents could hint at a problem.

Secondly, inspect the terms indexed in the database with `xapian-delve -a` . There will be a lot of terms, but can you spot problematic entries, such as a long chain of random characters?

Cheers,
Robert

------------------------------------------
Cyrus: Info
Permalink: https://cyrus.topicbox.com/groups/info/T420e77510eee80e1-Me1cc647a070f5d0b7558d223
Delivery options: https://cyrus.topicbox.com/groups/info/subscription
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.