[Dovecot] Large IMAP operations tend to timeout and causeproblems?

[Alen Salamun]

Hello Timo!

Well I just did try some large operations on alpha4 (CVS dated from
18.10.05).

System:

CentOS 4.2 x86_64
- Disk is AX100 EMC SAN connected with fibrechannel 2Gb with ext3 on it
- 1Gb RAM (Boot from SAN)
- Xeon 2.8Ghz

I tried copying 3800 mails from one folder to new one. WORKS!
Then I tried deleting this 3800 mails. WORKS!
Purging 3800 mails. WORKS!

Same thing with 5500 mails. WORKS!

So it looks like alpha4 helped to solve that problem! If I'll encoounter 
another index corruption, I'll let you know.

BR,
Alen

Timo Sirainen wrote:
> On Thu, 2005-10-20 at 10:54 +0200, Alen Salamun wrote:
>> Hello all!
>>
>> I have moved from Courier-IMAP to Dovecot some month ago. It is quite 
>> stable on everyday use (1.0alpha4) but it has constant problems with 
>> large IMAP/SSL operations. We use Maildir format.
>>
>> For example if I select 400 messages, mark them deleted and use "purge 
>> command" in Thunderbird (to actually delete them) TB starts to do 
>> something and runs and runs and runs....Then I get script timed out 
>> errror and I have to restart TB. That never happened with Courier-Imap.
>> Same thing happens for example moving 1000msgs to another folder or 
>> something like that.
> 
> So it's always about deleting messages from one mailbox?
> 
> And you have noticed this with the alpha4 pre-release too? Since it had
> one change that could have been related to this.
> 
>> Another thing is still that from time to time dovecot indexes get 
>> corrupted and users can't access some folders. Then I have to delete 
>> dovecot* in that Maildir folder and it works again.
> 
> Even if there are errors, Dovecot should usually fix the indexes itself
> without having to delete them manually. What errors exactly do you get?
> What filesystem do you use?