similar to: Startup errors

I am re-posting this message from last Friday, as I have seen no replies. It is critical that we address this problem, since until then we cannot put more than a few hundred users on our server. If Dovecot-auth TRULY needs to keep open a file descriptor for every single login session, we will have to start looking at alternatives. In short: Help! ----- Forwarded message from Steven F Siirila

We are running Dovecot 1.0-rc15 on Solaris 10 and are starting to ramp up our user count and ran into a file descriptor limit: Jan 1 19:38:54 <hostname> dovecot: IMAP(username): Newly created index file is corrupted: /home/username/.imap/foldername/dovecot.index.tmp Jan 1 19:38:54 <hostname> dovecot: IMAP(username): file mail-index.c: line 1806 (mail_index_move_to_memory): assertion

Here is another crash we've been seeing recently in rc15 on Solaris 10. (gdb) bt full #0 0xff1c12a4 in ?? () No symbol table info available. #1 0xff140040 in ?? () No symbol table info available. #2 0x000786a8 in t_buffer_alloc (size=688976) at data-stack.c:346 __PRETTY_FUNCTION__ = "file %s: line %" #3 0x00078190 in t_pop () at data-stack.c:149 frame_block = (struct

After upgrading a test server from rc15 to rc16, Postfix was no longer able to deliver mail via Dovecot deliver: Jan 5 11:00:59 testhost postfix/local[8873]: 497282EFE: to=<testuser at testhost>, relay=local, delay=1.2, delays=0.13/0.02/0/1, dsn=5.3.0, status=bounced (Command died with status 89: "/opt/dovecot/libexec/dovecot/deliver -f "$SENDER"") Should "status

We just experienced another crash in rc15. Here is the full backtrace: (gdb) bt full #0 0xff1c12a4 in .L423 () from /lib/libc.so.1 No symbol table info available. #1 0xff15fe20 in __wcsxfrm_std () from /lib/libc.so.1 No symbol table info available. #2 0xff140040 in getutline () from /lib/libc.so.1 No symbol table info available. #3 0x000786a8 in i_internal_panic_handler ( fmt=0x9ae80

For those running Dovecot on Solaris, what are your positive/negative experiences with building the software with CC vs. GCC? We are currently letting ./configure find our gcc and build with it, but wondered if anyone has any experience with Sun CC (does it even compile, for example). -- Steven F. Siirila Office: Lind Hall, Room 130B Internet Services E-mail: sfs at umn.edu Office of

Anyone running Dovecot on Solaris, requiring only SSL connections, and servicing thousands of simultaneous users? Our past attempts to do so have failed due to dovecot-auth using up all 256 of its file descriptors. I say 256 since we are limited to that by the 32-bit libraries. We are looking at either compiling Dovecot in 64-bit and letting dovecot-auth file descriptors accumulate (not ideal),

I hate to be a pest, but are there any revelations on file descriptor "overusage" by dovecot-auth? ----- Forwarded message from Steven F Siirila <sfs at tc.umn.edu> ----- Date: Sat, 20 Jan 2007 12:42:50 -0600 From: Steven F Siirila <sfs at tc.umn.edu> To: Dovecot Mailing List <dovecot at dovecot.org> User-Agent: Mutt/1.4.2.1i Subject: Re: [Dovecot] dovecot-auth file

I just upgraded our test server from rc6 to rc10... ISSUE #1 I noticed that we no longer are getting two "From_" lines. However, the one that was good (had the original sender e-mail address) is now gone. The one with "dovecot.deliver" remains. I attempted to add "-d test" to the mailbox_command line in main.cf as follows: mailbox_command =

We are having problems with the latest Dovecot 1.0 (RC15) regarding file descriptors > 255 being handed off to fdopen() via crypt() via PAM(?). This is a problem on Solaris 10 due to the fact that the stdio library does not support file descriptors > 255 (in order to remain binary-compatibile with older binaries). Here are the messages we run into if we get too many (~250?) concurrent

In dovecot-auth (rc15) we experienced problems when the file descriptors reached 256 (due to a Solaris fdopen() call described previously). The issue was that there was one file descriptor being used for every imap-login and pop3-login process (in our environment of SSL-only that means one for every active IMAP and POP connection). Mods were made (and appear in rc16) to close out the descriptor

I am having difficulting getting to www.dovecot.org and wiki.dovecot.org. My browser tells me it can't find them. Anyone know what's up? -- Steven F. Siirila Office: Lind Hall, Room 130B Internet Services E-mail: sfs at umn.edu Office of Information Technology Voice: (612) 626-0244 University of Minnesota Fax: (612) 626-7593

A few RC releases ago we had a problem where a user's restored home directory contained a .imap folder either from a previous RC release or perhaps it was a different server. In any event, restoration of the user caused problems since Dovecot choked on the restored .imap index files. Removing those index files cleared the problem. I'm wondering if this is still an issue, and therefore

Under Dovecot 1.0-rc27 on Solaris 10 we noticed this error today affecting one of our users repeatedly: Mar 28 14:02:01 myhost dovecot: IMAP(myuser): file mbox-sync-rewrite.c: line 423 (mbox_sync_read_and_move): assertion failed: (need_space == (uoff_t)-mails[idx].space) Mar 28 14:02:01 myhost dovecot: IMAP(myuser): Raw backtrace: 0x7ac54 -> 0x42cd4 -> 0x3e968 -> 0x3f54c -> 0x3f980

Okay, to convince myself that quota handling was working okay with Dovecot 1.0-rc27 on Solaris 10, I tested sending a message to an account that was purposely over-quota. The account in question had a vacation message set up. For the first test, there was no $HOME/.imap directory at all. I received the vacation message, but the message which was supposed to be delivered locally queued in Postfix

We are running Postfix 2.3.3 and Dovecot 1.0-RC27 on a Solaris 10 server. We use the Dovecot LDA for local delivery. Problems arise with looping mail whenever a user's .forward contains a reference to themselves. This most frequently occurs when a user sets up a vacation since it then puts this in their .forward file: \username, "|/usr/bin/vacation username" The vacation program

On Sun, Oct 08, 2006 at 10:25:53PM +0300, Timo Sirainen wrote: > On Fri, 2006-10-06 at 21:00 -0500, Steven F Siirila wrote: > > I am running Dovecot 1.0-rc6 along with dovecot-sieve from 2006-10-05 CVS. > > I am calling dovecot deliver from Postfix (main.cf): > > > > mailbox_command = /opt/dovecot/libexec/dovecot/deliver > > > > Upon switching to dovecot

I am doing some testing with Dovecot and wondered if it was normal for the deliver program to create index files for an Inbox (the test Inbox is over 7MB), but to not update the index files after the initial creation, even though messages continue to come into the mailbox. Note that during this testing, no IMAP or POP accesses are being made. Would it matter if a command-line program such as

When I configure Dovecot to conform to our directory hierarchy, it insists on adding 'dovecot/' to the prefixes I specify. Is there any way to inhibit this? My configure command is: ./configure --prefix=/opt/dovecot --sysconfdir=/etc/opt/dovecot \ --libexecdir=/opt/dovecot/libexec --localstatedir=/var/opt/dovecot \ --without-sql-drivers --with-ssl=openssl \

I don't recall if I mentioned this on the list, but when you are configured with mbox format, and a user has no Dovecot index files yet, and the config is such that mbox_min_index_size is set to non-zero, non-default (128 in my test case), dovecot deliver fails. Commenting out "mbox_min_index_size" fixed the problem, and the indexes then get created. This problems exists in RC10 --