Daemons started by systemd shouldn't fork into the background if
possible, because if multiple forked processes exist, systemd has
a hard time determining the main process ID.
In a memory constrained environment, the OOM killer may cause
ejabberd to exit without any trace. Because epmd keeps running,
systemd wouldn't notice the error condition, and as a result it
won't restart the server.
With ejabberd running in foreground, systemd is able to obtain the
correct exit code (137 in this case, instead of 0) and schedules a
restart. The administrator can then see what happend by looking at
systemctl status ejabberd.
This together with last history message is used by room_unused_* command
to determine for how long room was not used, this change allow us to skip
checking history, and works even when history is disabled.
This required adding hibernation time to data stored for room to be able
determine how old was a room, rooms that aren't stored with that
information will use node start time as timestamp for that.
This caused garbled text in some places in webadmin when using language
that used characters > 128.
Thanks to chengshq for noticing this and providing preliminary patch.
By doing this check_password that returned info about mismatched password
caused user_exists checks performed after that to return wrongly that
account doesn't exist.
Previously c2s was free to generate data to send in unlimited manner, and
just generate queue of messages that are waiting to be send. This could lead to
hitting timouts in stream management ack handling (if c2s generate lot of
packages, after which <r> request was sent, client could even not receive it
before timeout was triggered on server waiting for corresponding <a>).
This changes makes c2s process wait for data being sent to tcp socket
associated with websocket connection, which should help with this problem.
Since commit de91618070, it was no longer
possible to configure mod_register so that only account removal (i.e.,
no registration) is permitted. Revert to the previous behavior which
allows admins to freely configure account registration and removal via
separate access rules.
The docker local copies of DB initialization scripts were out of sync
with the master copies. Instead of updating local copies, adjust docker
compose to reference master copies directly.
Also change docker config to use docker volumes instead of mounted
directories for all DB data.
Correct Makefile clean targets
Fix a few more include() -> include_lib() for depedency includes
Use project_app_dirs to reference elixir app from rebar3 (lib_dirs is
no longer supported)
Sometimes we can observer combinations of subscription/ask that shouldn't
happen normally, but can be generated with api calls, let's try to handle
that gracefully instead of crashing.
