Panic: file dsync-brain-mailbox.c: line 814 ...

[Sander Lepik]

17.11.2016 11:01 Sander Lepik kirjutas:
> Hi!
> 
> We have 2 servers replicating each other and after upgrade to 2.2.26.0
> we are seeing this in the logs:
> 
> Primary server:
> 
> Nov 17 09:37:39 mailhost01 dovecot: dsync-server(user at domain.ee): Panic:
> file dsync-brain-mailbox.c: line 814 (dsync_brain_slave_recv_mailbox):
> assertion failed: (memcmp(dsync_box->mailbox_guid,
> local_dsync_box.mailbox_guid, sizeof(dsync_box->mailbox_guid)) == 0)
> Nov 17 09:37:39 mailhost01 dovecot: dsync-server(user at domain.ee): Error:
> Raw backtrace: /usr/lib/dovecot/libdovecot.so.0(+0x9438e)
> [0x7f3ccceb238e] -> /usr/lib/dovecot/libdovecot.so.0(+0x9447c)
> [0x7f3ccceb247c] -> /usr/lib/dovecot/libdovecot.so.0(i_fatal+0)
> [0x7f3ccce4ba4e] ->
> dovecot/doveadm-server(dsync_brain_slave_recv_mailbox+0x3d8)
> [0x7f3ccd8f66f8] -> dovecot/doveadm-server(dsync_brain_run+0x650)
> [0x7f3ccd8f4110] -> dovecot/doveadm-server(+0x4143b) [0x7f3ccd8f443b] ->
> dovecot/doveadm-server(+0x5735f) [0x7f3ccd90a35f] ->
> /usr/lib/dovecot/libdovecot.so.0(io_loop_call_io+0x4c) [0x7f3cccec6bdc]
> -> /usr/lib/dovecot/libdovecot.so.0(io_loop_handler_run_internal+0x10a)
> [0x7f3cccec809a] ->
> /usr/lib/dovecot/libdovecot.so.0(io_loop_handler_run+0x25)
> [0x7f3cccec6c65] -> /usr/lib/dovecot/libdovecot.so.0(io_loop_run+0x38)
> [0x7f3cccec6e08] -> dovecot/doveadm-server(+0x26b99) [0x7f3ccd8d9b99] ->
> dovecot/doveadm-server(+0x28efc) [0x7f3ccd8dbefc] ->
> dovecot/doveadm-server(+0x3daba) [0x7f3ccd8f0aba] ->
> /usr/lib/dovecot/libdovecot.so.0(io_loop_call_io+0x4c) [0x7f3cccec6bdc]
> -> /usr/lib/dovecot/libdovecot.so.0(io_loop_handler_run_internal+0x10a)
> [0x7f3cccec809a] ->
> /usr/lib/dovecot/libdovecot.so.0(io_loop_handler_run+0x25)
> [0x7f3cccec6c65] -> /usr/lib/dovecot/libdovecot.so.0(io_loop_run+0x38)
> [0x7f3cccec6e08] ->
> /usr/lib/dovecot/libdovecot.so.0(master_service_run+0x13)
> [0x7f3ccce51f53] -> dovecot/doveadm-server(main+0x19f) [0x7f3ccd8ccdef]
> -> /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf5)
> [0x7f3ccca94b45] -> dovecot/doveadm-server(+0x19ea6) [0x7f3ccd8ccea6]
> Nov 17 09:37:39 mailhost01 dovecot: dsync-server(user at domain.ee): Fatal:
> master: service(doveadm): child 42621 killed with signal 6 (core dumps
> disabled)

Looking at the logs more deeply we can see that it's pretty sure that it
was automatic full resync causing this error. But if that's the case
then replicator's status is giving wrong info.

It shows that full sync was successful and nothing failed - the process
crashed and there were no more retries in the logs - how can it be
successful?

-- 
Sander