- dovecot - dovecot.org

[dovecot] Failure in the presence of too many connections
by Mark E. Mallett 02 Apr '03

02 Apr '03

Had an event today where dovecot refused new connections and would not accept more until restarted, whereupon it worked for a few minutes and locked up again. Looking at the logs it appeared that it had run out of file descriptors. I increased the open files limit and started it again-- which worked but it only pushes off the problem. Does dovecot keep an open file descriptor for every imap or pop3 child? If so, this would argue more towards a tcpserver sort of control. If not, perhaps something is leaking.. -mm- PS: assume you saw recent RFCs 3501 and 3502 ?

3 14

[dovecot] dovecot-0.99.8.1 @ old redhat6.2 box
by Sebastian Pachuta 01 Apr '03

01 Apr '03

Hi I have some odd problem with running dovecot. Program dies with message in maillog: Mar 31 21:04:02 test-box dovecot: Dovecot starting up Mar 31 21:04:04 test-box dovecot: execv(imap-login) failed: Resource temporarily unavailable Mar 31 21:04:04 test-box dovecot: execv(imap-login) failed: Resource temporarily unavailable Mar 31 21:04:04 test-box dovecot: Login process died too early - shutting down Mar 31 21:04:04 test-box dovecot: child 28870 (login) returned error 84 (exec() failed) Mar 31 21:04:04 test-box dovecot: child 28869 (login) returned error 84 (exec() failed) Mar 31 21:04:04 test-box dovecot: execv(imap-login) failed: Resource temporarily unavailable System is redhat6.2 (with all updates), kernel is 2.2.25 (with openwall patch). Under rh7.3 dovecot launches fine.

2 4

[dovecot] NFS mounted IMAP dirs
by Maikel Verheijen 31 Mar '03

31 Mar '03

Hi list, I am trying to build an email system that is using multiple frontends, to a clustered backend using nfs. The backend mounts his filesystems from a central storage, and will be running active-active. While reading through the dovecot documentation, I read that I could do without indexes (well: use memory indexes), and could then be able to run dovecot over NFS. Well: that was until friday, when I was reading a little bit better through nfs.txt. :) My problem is now that dovecot writes it's subscriptions to the mailstore. Since I want as little difference between my frondends as possible, (and as static as possible), I would love to see dovecot handle subscriptions in a database, or perhaps a nfs-safe file. Since I am only a junior programmer (my code is really not far from "Hello world", although I am trying to find some time to teach myself more), I might not be able to be of much assistance to you all in coding stuff, but I am willing to help out where-ever possible. Can anyone help me solve my problem in any way, and how can I contribute as well? Thank you all, Maikel Verheijen.

2 1

[dovecot] FETCH BODY[part]<partial>
by Jesse Peterson 29 Mar '03

29 Mar '03

2003-03-24 12:59 Timo Sirainen <tss(a)iki.fi> This CVS update didn't actually seem to fix the problem. I still get the problem as before, like this capture: 10 UID FETCH 60 BODY.PEEK[]<8192.1915> * 60 FETCH (UID 60 BODY[]<8192> {1915} ) 10 OK Fetch completed. Not quite 1,915 bytes =). Perhaps the "[]" (without a part) is confusing it? But still, it grabs the first chunk correctly. At any rate, any input would be great.

2 6

[dovecot] Dovecot and Quotas
by Mike Machado 26 Mar '03

26 Mar '03

Is there anything that can be done so that when users IMAP to their inbox and have reached their filesystem quota, that they can still DELETE their mail? They can login, but when they go to delete/expunge this is the error I get in syslog. I have tried mail_read_mmaped to yes and no, and same result. Mar 25 14:54:27 penguin1 imap(bob): file_set_size() failed with index data file /home/bob/mail/.imap/INBOX/.imap.index.data: Disk quota exceededMar 25 14:54:27 penguin1 imap(bob): mmap_istream.madvise(): Function not implemented Mar 25 14:54:28 penguin1 last message repeated 65 times When using UW IMAP, sometimes users can delete mail and other times they cannot. I am desperately seeking a solution to have customers be able to delete their own mail when their box is full and not have to call our support center. Any suggestions? Thanks. Debian with kernel 2.4.18. -- Mike Machado mike(a)innercite.com InnerCite Inc. Engineering Director / CTO

5 6

[dovecot] Re: mailbox problem
by Maikel Verheijen 26 Mar '03

26 Mar '03

> Hmm. Unless you're using one UID for everyone, those > directories would have be to be created as root. Also > test.com should be created with different owner as blah and > blah/Maildir. Maybe I could add a new setting for this, but > I'm not sure how to do it without making it too kludgy. There > would have to be separation of where user directory structure > starts, and uid/gid settings what to use for created > directories before them. Something like: I do use 1 UID and 1 GID for everyone. Would that be more simple? > mail_global_dir = /var/mail/t/te/%d > mail_global_uid = root > mail_global_gid = mail > mail_user_dir = /var/mail/t/te/%d/%n > > Would be better if the mail_user_dir didn't have to be set at > all since it's already in mail_default_env, but the user > directory has to be created while still root, and parsing > mail_default_env there is kludgy too. Or maybe: > > mail_global_dir = /var/mail/t/te/%d/%n > > And the last directory would always be assumed to be user > directory. That would prevent creating /var/mail-like shared > directories though, but I guess there's no need to create those.. Maybe there could be a "toggle" to do or don't create the dirs, so you don't need the mail_global_dir setting for "normal" use. My idea was to check the full path to the mailbox, and to create the nonexisting part if necessary. At that point you DO have the full path to the mailbox/maildir, and you can create it, right? Kind regards, Maikel Verheijen.

2 1

[dovecot] inetd/xinetd/tcpserver support
by Timo Sirainen 26 Mar '03

26 Mar '03

I was just thinking how they could be easily supported. This would work, right? : imap stream tcp nowait root /usr/sbin/tcpd /usr/local/libexec/dovecot/imap-login imaps stream tcp nowait root /usr/sbin/tcpd /usr/local/libexec/dovecot/imap-login --ssl imap-login would try to connect to master process using some named socket. If it couldn't, it would create the master process itself. Master process would work as usual (executes auth and imap processes), except it wouldn't be executing login processes. This wouldn't require much code changing, and it would still be using all the same privilege separations as the standalone version so it would be just a secure. Only thing I'm wondering is if any of the TCP wrappers care about the created child processes? The master process would have to stay alive after the connection that created it dies. I guess I'll implement this soon and try if it works.

7 20

[dovecot] mailbox problem
by Maikel Verheijen 26 Mar '03

26 Mar '03

Hi list! Since a week I am testing with dovecot, and I must say, I like it a lot :) Keep up the good work :) I do have one questing regarding dovecot: If I login to a mailbox, and the mailbox does NOT exist, and the parent directory does not exist either, I will get disconnected, and dovecot claims it cannot create the directory. (eg: my user blah(a)test.com has mail directory /var/mail/t/te/test.com/blah/Maildir, and all directories up to /var/mail/t/te/ exist, dovecot does not create the directory test.com/ and on.) Is this by design, or should this be created by dovecot? The problem is solved as soon as the user receives his first email, since my MTA creates the directories... Thank you, Maikel Verheijen

2 1

[dovecot] Help
by Barker, Ron, vpe 26 Mar '03

26 Mar '03

Hi to all, We are a bit of newbes to imap etc, so may be we just not doing it right. Nevertheless, We have compiled dovecot to run under Solaris 8/9 for IPv6 test. We are going crazy. After a successful compilation, the troubles really began. Solaris 9 1. Every´time we reboot- the directory /var/run/dovecot/auth is gone ???? 2. After the dir is created manually, the server starts. However!!! 1. A telnet to host 143 connect to the server fine but no commands are accepted. The server simply returns "bad command" 2. If we try to connect via a e-mail client to the server - we get a login refused - which I suspect is simply the "bad command" error being interpreted as such. We have tried every concivable combination of conf items in the conf file but nothing. Solaris8 Just does not start. Any help apprecialted Ron -- Dr. Ronald Barker Senior Technology Manager Vodafone Group Research & Development - D Vodafone Pilotentwicklung GmbH Chiemgaustrasse 116 81549 Munich Germany Fon: + 49 (89) 95 410-0 Fax: + 49 (89) 95 410-111 mailto:ron.barker@v-pe.de http://www.vodafone-rnd.com <<Barker, Ron, vpe.vcf>>

2 1

[dovecot] dovecot+Solaris 9
by Barker, Ron, vpe 26 Mar '03

26 Mar '03

Does the /dev/urandom issue apply to Solaris 9 as well??? Thanks Ron -- Dr. Ronald Barker Senior Technology Manager Vodafone Group Research & Development - D Vodafone Pilotentwicklung GmbH Chiemgaustrasse 116 81549 Munich Germany Fon: + 49 (89) 95 410-0 Fax: + 49 (89) 95 410-111 mailto:ron.barker@v-pe.de http://www.vodafone-rnd.com <<Barker, Ron, vpe.vcf>>

2 1