Thanks for hint. So I checked with strace and its output is here:

http://www.volny.cz/vykupitel/Pictures/mail/dovecot.err

I think that the most interesting part is on the end. It segfault on the parameter mail_full_filesystem_access = no.

. . . uname({sys="Linux", node="mail.valbek.cz", ...}) = 0 rt_sigaction(SIGINT, {0x49b680, [], SA_RESTORER|SA_SIGINFO, 0x3129a302d0}, NULL, 8) = 0 pipe([4, 5]) = 0 fcntl(4, F_GETFL) = 0 (flags O_RDONLY) fcntl(4, F_SETFL, O_RDONLY|O_NONBLOCK) = 0 fcntl(5, F_GETFL) = 0x1 (flags O_WRONLY) fcntl(5, F_SETFL, O_WRONLY|O_NONBLOCK) = 0 fcntl(4, F_GETFD) = 0 fcntl(4, F_SETFD, FD_CLOEXEC) = 0 fcntl(5, F_GETFD) = 0 fcntl(5, F_SETFD, FD_CLOEXEC) = 0 epoll_create(128) = 6 fcntl(6, F_GETFD) = 0 fcntl(6, F_SETFD, FD_CLOEXEC) = 0 epoll_ctl(6, EPOLL_CTL_ADD, 4, {EPOLLIN|EPOLLPRI|EPOLLERR|EPOLLHUP, {u32=200799440, u64=200799440}}) = 0 rt_sigaction(SIGTERM, {0x49b680, [], SA_RESTORER|SA_SIGINFO, 0x3129a302d0}, NULL, 8) = 0 rt_sigaction(SIGPIPE, {0x1, [], SA_RESTORER|SA_RESTART, 0x3129a302d0}, NULL, 8) = 0 rt_sigaction(SIGALRM, {0x49b480, [], SA_RESTORER|SA_SIGINFO, 0x3129a302d0}, NULL, 8) = 0 rt_sigaction(SIGXFSZ, {0x1, [], SA_RESTORER|SA_RESTART, 0x3129a302d0}, NULL, 8) = 0 geteuid() = 0 open("/etc/dovecot.conf", O_RDONLY) = 8 fstat(8, {st_mode=S_IFREG|0644, st_size=3618, ...}) = 0 pread(8, "mail_full_filesystem_access = no"..., 4096, 0) = 3618 --- SIGSEGV (Segmentation fault) @ 0 (0) --- +++ killed by SIGSEGV +++

Best Regards, Ladislav

Dne 30.7.2010 18:04, Pascal Volk napsal(a):

On 07/30/2010 05:58 PM Pašek Ladislav wrote:

...

According to my previous email, is it possible to somehow debug delivery process to find out why is this happening?

Command died with signal 11: "/usr/libexec/dovecot/deliver"

Thanks in Advance

Ladislav

See http://dovecot.org/bugreport.html Note: you can also execute deliver on the commandline deliver -d recipi@ent -f enevelope@sender< test_message_file See http://wiki.dovecot.org/LDA#Parameters for more details.

Regards, Pascal

Thanks for hint. This is output from gdb bt full but I have no clue what to find in it: #0 0x0000003129a32ce2 in unsetenv () from /lib64/libc.so.6 No symbol table info available. #1 0x000000000041a078 in config_file_init (argc=<value optimized out>, argv=<value optimized out>) at deliver.c:497 No locals. #2 main (argc=<value optimized out>, argv=<value optimized out>) at deliver.c:1062 _data_stack_cur_id = 2 config_path = 0x4aa3e6 "/etc/dovecot.conf" mailbox = 0x4b32f5 "INBOX" auth_socket = <value optimized out> home = <value optimized out> destaddr = 0x0 user = 0x7fff50d9272b "pasek@valbek.cz" errstr = <value optimized out> path = 0x0 orig_user = <value optimized out> extra_fields = {arr = {buffer = 0x0, element_size = 0}, v = 0x0, v_modifiable = 0x0} mail_user = <value optimized out> raw_mail_user = <value optimized out> raw_ns = <value optimized out> e = <value optimized out> box = <value optimized out> raw_box = <value optimized out> input = <value optimized out> t = <value optimized out> headers_ctx = <value optimized out> mail = <value optimized out> cwd = "\360>\265s\255*\000\000ù`)1\000\000\000\001\000\000\000\000\000\000\000\360\071\265s\255*\000\000\020\004\331P\377\177\000\000\000\000\000\000\000\000\000\000\353\214@41\000\000\000\266\270`)1\000\000\000\001\000\000\000\000\000\000\000\030\065\265s\255*\000\000P\b\331P\377\177\000\000\000\000\000\000\000\000\000\000\300\214@41\000\000\000\266\270`)1\000\000\000\001\000\000\000\000\000\000\000H0\265s\255*\000\000\200\b\331P\377\177\000\000\000\000\000\000\000\000\000\000\260\214@41\000\000\000\266\270`)1\000\000\000\001\000\000\000\377\177\000\000\200+\265s\255*\000\000\260\b\331P\377\177\000\000\000\000\000\000\000\000\000\000\243\214@41\000\000\000\266\270`)1\000\000\000\001\000\000\000\000\000\000\000\250&\265s\255*\000\000\340\b\331P\377\177\000\000\000\000\000\000\000\000\000\000\217\214@41\000\000\000\266\270`"... process_euid = 0 stderr_rejection = false keep_environment = false user_auth = true mtime = <value optimized out> i = <value optimized out> ret = <value optimized out> userdb_pool = 0x0 str = <value optimized out> error = MAIL_ERROR_NONE Pátek, 30 Červenec, 2010 21:43 CEST, Timo Sirainen napsal:

On 30.7.2010, at 18.44, Ladislav Pašek wrote:

...

Thanks for hint. So I checked with strace and its output is here:

Strace output isn't very helpful for crashes. gdb backtrace would show exactly where it crashes. If you can crash it by running

deliver ..params.. < mail-file

then easiest way to get backtrace is to run it like:

gdb --args deliver ..params.. run < mail-file bt full

Yes it was it, I 've got on the first line mail_full_filesystem_access = no. Now is it working.

Thanks a lot Timo

Best Regards

Lada

Úterý, 03 Srpen, 2010 23:14 CEST, Timo Sirainen tss@iki.fi napsal:

On 3.8.2010, at 22.09, Pašek Ladislav wrote:

...

Thanks for hint. This is output from gdb bt full but I have no clue what to find in it:

#0 0x0000003129a32ce2 in unsetenv () from /lib64/libc.so.6

Oh. This again. Just make sure the first setting in your dovecot.conf isn't "something=no". Move it down in the file.