[ejabberd] Server allocating all available ram, then stops at 100% CPU

Peter Schwindt ejabberd at schwindt-net.de
Mon Nov 11 10:45:32 MSK 2013


Hi

we're experiencing some unfunny behaviour at jabber.ccc.de these days:

After some time running the ejabberd (2.1.13) beam.smp (erlang 15b1)
process is seen to run at > 150% CPU, allocating more and more ram and
afterwards it stops accepting and forwarding packets and runs at 100% -
which therefor isn't usable so we have to kill the process and restart.

Having a look at the last log entries in ejabberd.log there are quite a
few of these:

=ERROR REPORT==== 2013-11-11 07:35:27 ===
E(<0.487.0>:gen_iq_handler:118) : {badarg,

[{erlang,phash,[{1384,151727,368591},0],[]},
                                    {ejabberd_odbc_sup,get_random_pid,1,
                                     [{file,"ejabberd_odbc_sup.erl"},
                                      {line,108}]},
                                    {ejabberd_odbc,sql_call,2,

[{file,"ejabberd_odbc.erl"},{line,124}]},
                                    {mod_roster,read_roster_version,3,
                                     [{file,"mod_roster.erl"},{line,201}]},
                                    {mod_roster,roster_version,2,
                                     [{file,"mod_roster.erl"},{line,180}]},
                                    {mod_roster,push_item,4,
                                     [{file,"mod_roster.erl"},{line,555}]},
                                    {mod_roster,process_item_set,3,
                                     [{file,"mod_roster.erl"},{line,481}]},
                                    {lists,foreach,2,
                                     [{file,"lists.erl"},{line,1262}]}]}


Any ideas? Is there another possibility to look for the reason of these
problems?

Thanks,
Peter


More information about the ejabberd mailing list