By default, queries and stores the total memory usage in
AdminModel every 5 seconds, and caches the last 100 such values.
Both cache size and interval can be changed by simple commands
from the clients.
Change-Id: I86cf8228d0129dc8aab0a03856c12dfeb240b169
Well, it was already there.
e7f75a3e56 was though correct
introduced it again. Remove duplicate such call.
Change-Id: Iddacbee53aaec58340db489e7830af75354aee34
It is desirable to have colored logs when running loolwsd
in a terminal, but not redirecting its output to a file.
Outputting to a terminal is now detected and colored logs
are automatically enabled.
To force colored logs in files as well, define
LOOL_LOGCOLOR in the environ. The output color codes
can then be processed using, f.e., `less -r`.
Change-Id: I09fbee4441f210d814ac5ad23dd99d1c33b560b7
Reviewed-on: https://gerrit.libreoffice.org/23080
Reviewed-by: Ashod Nakashian <ashnakash@gmail.com>
Tested-by: Ashod Nakashian <ashnakash@gmail.com>
... for which they want to be notified.
Needed to add some missing header files to Storage and Auth class
because of conflict raised when LOOLWSD.hpp is included in
Admin.cpp
Change-Id: Ia1c8ed82f8cd979eaf93267ae5dfa347acf895f4
Further changes/refactoring to make it possible:
* Add broker pid to Admin class
* Move getMemoryUsage for process to Util
* Change variable name to accurately reflect *active* items
_nViews -> _nActiveViews, etc.
Change-Id: I4c9206c49ab829b73ebfe226874bfbbcc8f95342
Reviewed-on: https://gerrit.libreoffice.org/22989
Reviewed-by: Tor Lillqvist <tml@collabora.com>
Tested-by: Tor Lillqvist <tml@collabora.com>
Otherwise the process could terminate while admin sessions are still
being served.
Change-Id: Id91d0989e264e8294827e8e2ad8dd2d44b6006ec
Reviewed-on: https://gerrit.libreoffice.org/23021
Reviewed-by: Tor Lillqvist <tml@collabora.com>
Tested-by: Tor Lillqvist <tml@collabora.com>
Much simpler.
Also, don't duplicate the code informing that LD_BIND_NOW or
LOK_VIEW_CALLBACK are not set. Not that I understand why we need to
inform about that? If the "normal" thing should be that they are set,
why don't make it so by default then?
I occasionally saw a deadlock when running 'make check' where one
thread holds the ChildProcessSession::Mutex and wants the _mutex for a
Document, while another thread holds that _mutex and wants the
Mutex. In particular, it is the Document::onUnload() that wants the
_mutex. So avoid the deadlock by having Document::onUnload() first
take the ChildProcessSession::Mutex.
The callbacks from documentLoad() are made in the same thread.
Sure, as such it is not a good thing to use recursive mutexes. If we
switch back to non-recursive mutexes, we will have to stop taking the
lock in callbacks from documentLoad(), i.e. make sure we know those
functions aren't used elsewhere, in places where a lock would be
needed. Or something.
If a client session closes just after sending a load message to load a
document, and another session then fairly immediately connects and
sends a load message for the same document, the latter session gets
handled by the same kit process. Also, the same Document object is
apparently used. In that kit process, the first documentLoad() might
easily still be in progress. The handler for the new session still
calls onLoad(), too, and as the first onLoad() had dropped the lock
for the duration of the documentLoad() call, the new onLoad can take
the lock and call documentLoad(), too, while the first documentLoad()
call in the other thread still is in progress. This leads to
interesting problems.
It is loolwsd that spawns loolbroker so we control what arguments it
gets, so no need to give verbose errors if our own code is
inconsistent. That is what assert() is for.
Admin web sessions are added as subscribers to AdminModel. Live
notification fill up the AdminModel, and notifies to
subscribers, if present any. AdminModel can also be queried to
fetch any previous data since the start of the server including
expired documents/views with timestamps for analysis.
There is lot of stuff that can be added in future. This commit
just lays the foundation of appropriate classes.
Change-Id: Ifcf6c2896ef46b33935802e79cd28240fd4f980e
Reviewed-on: https://gerrit.libreoffice.org/22869
Reviewed-by: Tor Lillqvist <tml@collabora.com>
Tested-by: Tor Lillqvist <tml@collabora.com>
As a test, add command to fetch documents from AdminModel.
Change-Id: I3cb7097ba7dde049f3b2478fe7b6b6c309da1d92
Reviewed-on: https://gerrit.libreoffice.org/22781
Reviewed-by: Tor Lillqvist <tml@collabora.com>
Tested-by: Tor Lillqvist <tml@collabora.com>
Turns out the callbacks from documentLoad() are in general done on
another thread, not necessarily the same one that did the
documentLoad() call. Not sure why it happened to be the same thread
for me, but oh well. Need to fix the problem described in
3671abf89b in some other way then.
This reverts commit 2fab757462.
The callbacks from documentLoad() are made in the same thread.
Sure, as such it is not a good thing to use recursive mutexes. If we
switch back to non-recursive mutexes, we will have to stop taking the
lock in callbacks from documentLoad(), i.e. make sure we know those
functions aren't used elsewhere, in places where a lock would be
needed. Or something.
... and Admin and AdminModel containing all the required data
that we need to expose to Admin panel.
Admin processor will keep listening to any data on this
notification pipe and update AdminModel accordingly.
Change-Id: I0dd6f07ae60158733c34d17f53a35def70600513
Reviewed-on: https://gerrit.libreoffice.org/22780
Reviewed-by: Ashod Nakashian <ashnakash@gmail.com>
Tested-by: Ashod Nakashian <ashnakash@gmail.com>
It does quite a lot of work that can produce very many lines of log,
so it is good to be able to quickly find when that is done when
perusing a log file.
Not needed after all. It was a red herring. The device files work fine
even if not owned by root:root and with mode 664. The actual problem
was that I used a file system mounted with nodev when testing loolwsd.
This reverts commit 509314d559
That is probably what was the intent. As originally written, in case
the function encountered partial writers, and had to do several
write() calls, only the number of bytes written by the last one was
returned.
Luckily the actual return value of writeFIFO() is not used
anywhere. It is just tested for being negative.
Still there is the problem that if at first one or several write()
calls succeed but don't write the whole buffer, and then a write()
fails, the caller has no way to know that the buffer has been
partially written. But that is hopefully highly theoretical and there
is no sane way to handle such a situation anyway.
The getChildStatus() and getSignalStatus() functions returned the
latter, still the returned values were compared against
Poco::Util::Application::EXIT_OK. (Sure, both Poco's
Application::EXIT_SUCCESS and stdlib.h's EXIT_OK are zero, but that
doesn't mean one should mix them up.)
Also add two comments pondering the meaning of the code.
We are not interested in the variable being assigned an incremented
value. Its value is not used any more. We are just interested in the
value of the variable plus one. Using pre-increment gives the wrong
impression.
Sure, this is nit-picking.
So don't give it any then.
Remove the --uid option and related attempts to handle running loolwsd
under sudo, to be able to debug it. Now with loolwsd not having
capabilities, it should work fine to just run it under a debugger
normally. (For the loolbroker and loolkit processes, attaching to an
already started process is the way to debug.)
One has to love arbitrary retry counts and timeouts. Loading the
password-protected.ods in a loolkit process, with correct password
provided, takes 12 seconds on my machine. I think this slowness is
because the NSS code LO uses to do crypto wants to initialize its
crypto goodness in various ways that don't work so well inside a
chroot jail. Presumably it uses some wait with timeout when attempting
to do something which doesn't succeed. For instance it tries to run
netstat -in. (In an interactive LibreOffice the doc loads fairly
instantly.) Oh well.
It does three separate things, and the first two intentionally result
in errors, and the server probably disconnects after errors. Not
sure. This is horrible, horrible.
Having a separate "disconnect" message is a disgrace. There should be
no need for it. WebSocket has a perfectly fine graceful disconnect
mechanism already (CLOSE frames). The code needs to be prepared to
receive a CLOSE frame at any time. The code also needs to be prepared
for the underlying socket being bluntly closed by the other end,
without sending any WebSocket CLOSE frame. The only sane thing is to
handle a "disconnect" message in the same way as those situations
anyway, so why is it needed?
Sort #includes and using statements. Use 'using' consistently for all
Poco:: types. (I am not 100% convinced that using 'using' like done
here in loolwsd was a good idea after all. But at least let's be
consistent now that we do use it.)
No need to pass the value of a variable, initialised much earlier, to
a system call when one can pass the required constant value as
such. Much clearer.
