how to profiling imap process with valgrind
Hi, I want to profile runtime performance of imap and pop3 processes. There processes are forked by dovecot master process. I am wondering if there is a way I can profile these processes with valgrind. I tried service = /bin/valgrind /bin/imap but this did not work. Thanks,
[Dovecot] Feature Request: handle mail connections of the same user from a same IP using just one mail process
Hi, Dovecot Wiki states that mail process (IMAP/POP3) is able to handle multiple client connections as the same time to save CPU and memory resources (client_limit 1). Although this approach is not recommended due to latency and probably security issues, it does help increase the overall capacity of a mail server. Is it possible (or planned) to add a feature such that mail client connections of the same user from a same IP are directed to a same mail process? In most cases, such connections come from a same mail client (e.g. in Thunderbird, each mailbox is a connection). The end user (human being) can work on one mailbox at a time, the latency will be non-obvious. And since all data being processed belong to the same user, security is a non-issue. Thanks,
Re: [Dovecot] versioning of mailbox index files
My current approach is to introduce a index version file .dovecot.index.version in each mailbox and hooks mailbox_allocated event via my own plugin. If the version file does not exist or exist but with version number which need to be upgraded, index files in that mailbox are removed. I also tried to hook mailbox_opened() event and found that event was too late (index was loaded already). The side effect of hooking mailbox_allocated is that the event fires not only by 'select' command. 'list' command can also trigger a lot of mailbox_allocated events. Are there any better hook points I can use for this purpose? Thanks,
[Dovecot] versioning of mailbox index files
Hi, I am wondering if dovecot assigns version information in mailbox index files so that by which version of dovecot the index files are created can be determined. Our existing mail service used dovecot 2.2.4 with lots of active mail users. When upgraded to 2.2.10, imap processes aborted all the time due to the corrupted index files (the problem was fixed in dovecot 2.2.5). After investigation, I found that this annoying problem could be avoid by deleting index files before opening a mailbox. Now the problem is how I could know if the index files are created by a version newer than 2.2.4. Thanks, Morrison
Re: [Dovecot] imap abort after upgrade from 2.2.4 to 2.2.7
When I did upgrade, 2.2.7 was the latest release. When I ran into the problem, I checked back and found 2.2.8 and 2.2.9 were released shortly one after another. It seemed these releases were sort of buggy. I would like to know if what I described in my previous email was an known issue and was addressed before upgrade to 2.2.10. We are using dovecot as our internal mail service. So stable is most important to us. Regards,
[Dovecot] imap abort after upgrade from 2.2.4 to 2.2.7
Hi, I used to use dovecot 2.2.4. I saw lots of warnings saying 'Extension header update points outside header size'. I upgraded to dovecot 2.2.7, based on the discussion in mail list. Now, I see lots of imap aborts around: src/lib-index/mail-index-transaction-export.c:203: i_assert(u32.offset + u32.size = ext_hdr_size); I also noticed, in some mailbox, imap process kept abort until all index files in the mailbox were manually deleted. Did someone run into the same problem and any solution to this? Thanks,
[Dovecot] SSL/TLS handshake stays forever without timeout
Hi, I am a system admin and I am evaluating using dovecot as our email server. In my test, I found that if I telneted to 993 port and did not do anything or I telneted to 143 port, sent starttls command and then did not do anything, the connection stayed forever without timeout. This will make our mail server vulnerable to DOS attack. I dig into dovecot Wiki and did not find any solution. This seems to me that dovecot does not handle SSL/TLS handshake timeout. I am wondering if this is a known issue and will be fixed in near future. Thanks,
Re: [Dovecot] [courier-dovecot-migrate.pl] - No such file or directory at courier-dovecot-migrate.pl line 230
On Mon, 23 Apr 2007 15:25:57 -0400 Charles Marcus [EMAIL PROTECTED] wrote: If you have a policy of this nature, then it is incumbent upon you to post via an address that doesn't add it. It's clearly obvious that the disclaimer is meaningless lawyer-speak in the context of a mailing list post. The OP should arrange to use another mail posting route such as a webmail address, this sort of thing is poor manners and should be resisted before this crap buries the usefulness of mail in pointless drivel. While I agree with the sentiment, I'm not sure how practical it is... If I see this sort of thing after I've told someone they should not be doing it, they go in the kill file. That's then one more person who won't be helped, and others who won't see useful suggestions if I have them, I only have so much time to help and doing so for someone who can't see the clash between their disclaimer and where they're posting is time I don't have at all. Kind of like complaining about top-posting or HTML email... it's there, and it isn't going anywhere. When you let it bother you - it only bothers *you*... Easy, to posters and html-only mail goes straight in the bin. And if you think I'm a net-Nazi, I can introduce you to people who can peel the paint off a tank at 100 yards just by opening their mouths to rant :) -- Brian Morrison bdm at fenrir dot org dot uk Arguing with an engineer is like wrestling with a pig in the mud; after a while you realize you are muddy and the pig is enjoying it. GnuPG key ID DE32E5C5 - http://wwwkeys.uk.pgp.net/pgpnet/wwwkeys.html
[Dovecot] Test
Hi Georgie Just testing! -- Brian Morrison bdm at fenrir dot org dot uk Arguing with an engineer is like wrestling with a pig in the mud; after a while you realize you are muddy and the pig is enjoying it. GnuPG key ID DE32E5C5 - http://wwwkeys.uk.pgp.net/pgpnet/wwwkeys.html
Re: [Dovecot] Released 1.0.rc31
On Mon, 9 Apr 2007 15:16:49 -0500 Steven F Siirila [EMAIL PROTECTED] wrote: RC31 is screaming for version 1 to be released: 31 == 1 binary. :) You're suggesting that 1.0 will be full of variable overflow bugs then? ;-) -- Brian Morrison bdm at fenrir dot org dot uk Arguing with an engineer is like wrestling with a pig in the mud; after a while you realize you are muddy and the pig is enjoying it. GnuPG key ID DE32E5C5 - http://wwwkeys.uk.pgp.net/pgpnet/wwwkeys.html
