Searching via search bar takes very long time in large mailboxes

[thstyl2000]

Steps to reproduce

1. Go to a mailbox containing a lot of emails
2. Search for something using the search bar

Expected behavior

You get a results in an acceptable amount of time.

Actual behavior

You get no results.

Mail app version

4.0.7

Nextcloud version

30.0.2

Mailserver or service

dovecot (imap)

Operating system

freebsd

PHP engine version

PHP 8.2

Nextcloud memory caching

'memcache.local' => '\OC\Memcache\APCu', 'memcache.locking' => '\OC\Memcache\Redis',

Web server

Nginx

Database

MariaDB

Additional info

The problem is because it also defaults to searching in the mail body (which is probably very expensive). Searching via the "Extra Settings" window (next to the bar) and using only one field returns results extremely quickly also in very very large mailboxes.

[SebastianKrupinski]

Hi @thstyl2000

We don't actual do a full body search, the body is only pulled for messages that have NOT been processed by our background worker, so that the structure and preview text can be analyzed.

The background job is "OCA\Mail\BackgroundJob\PreviewEnhancementProcessingJob" if you run the cron.php this should no longer pull the body of the message. Also after running a search once those messages should no longer be pulled again.

[thstyl2000]

I've tested this in my "Archive" mailbox which is huge, but I suppose that most of the message bodies must have already been pulled, since I've read all emails before archiving them. So what I'm inferring from your comment is that the background job has not been doing its job properly, or do I understand it wrong....?

[ChristophWurst]

Dear @thstyl2000,
Thank you very much for your report.

The app indeed searches in bodies, but it is opt-in due to the potential performance implications. See #8050 for details.
If you turn off body search globally, do you get faster results?

Prepocessing is irrelevant for search. The app can search for subject, flags and senders/recipients in the database. For searches in the message body the IMAP server is always contacted.

I hope this information allows you to narrow down where the bottleneck lies.