The logging of syslog messages is now done outside of the log_write
function. This removes the need to pass the prefix length as an
argument. With this change, the log_write function only writes MaxScale
log messages.
Also reworked the timestamp generation functions so that they return a
string. This greatly simplifies the message writing.
Removed skygw_utils and relate files along with the old log manager
code. Also removed file flushing due to it being redundant; messages are
written to the file immediately. Adjusted tests to accommodate this
change.
The log manager now uses the logger type for logging. Removed some of the
code that depended on the renamed functions to make it compile. The next
step is to remove all unused code in the log manager.
The interface defines a very simple logging API. The main logger,
FileLogger, implements a very basic logger that writes to a file. The
StdoutLogger implements a logger that simply writes to stdout.
The logging to shm could be implemented quite easily but the usefulness of
it is very questionable and it has the problem of losing log files on a
server shutdown. A ring buffer of INFO messages would be more useful and
would provide a way to limit the amount of memory used for it.
The feature was rarely used and was only useful in extremely rare
cases. The functionality can still be emulated, if for some reason needed,
by pointing `logdir` to `/dev/shm` or another tmpfs mount.
- Cannot be supported with std::thread.
- Unlikely that the default size (8MB) would ever be too small and
if it is, there is some problem.
- The stack size can be specified using 'ulimit -s' before starting
MaxScale.
The dummy session was unnecessarily modified by multiple threads at the
same time. The state of the dummy session must not change once it's been
set at startup.
The custom random number generator can now be replaced with a C++11
RNG. This greatly improves the reliability and trustworthiness of it.
In addition to this, the conversion of the RNG to a thread-local object
removes the race condition that was present in the previous
implementation. It also theoretically improves performance by a tiny bit.
If a module is not found, a proper error message is now given to the
user. Other system errors are also printed if errno is set. Dynamic
library errors aren't printed as they are apparently reset after a call to
`dlerror`, at least according to documentation.
It does not save much in characters compared to std::function<void ()>
and it's a bit misleading as not just any callable object will do, but
only one that takes no argument and returns void.
In preparation for Worker::post() to be renamed to Worker:execute().
The concept of _posting_ will be reserved to mean the transfer of
something over the message queue to the worker for processing and
nothing else.
This is the first step in some cleanup of the Worker interface.
The execution mode must now be explicitly specified, but that is
just a temporary step. Further down the road, _posting_ will
*always* mean via the message loop while _executing_ will optionally
and by default mean direct execution if the calling thread is that
of the worker.
The 'events' and 'script' config values were defined for every monitor.
Removed the extra definitions and moved the variables to MXS_MONITOR.
MariaDBMonitor was printing config values a second time, they are
already printed by the caller.
Moved the events enum definition to the internal header since it's no longer
required by modules.
Added a default config setting "all" to 'events' to clarify that it enables
all events.
The ssl parameters were defined as strings even thought they were actually
enums. The events parameter was also a string even though it was an enum.
Also added the missing "all" value to the events enum. This fixes the
regression of scripts not being launched on all events by default.
Moved the definition of the default version string where it should be and
removed the empty value check.
Took the size type validity check function into use in runtime
configuration. This fixes the problem with zero not being accepted as a
valid size for query_classifier_cache_size.
The get_suffixed_size function is now exposed in the internal config
header and it also checks for the validity of the size types.
Took the new function into use and added the appropriate error messages.
If a service has filters, they must first be removed before the service
can be destroyed. This is not a functional requirement but it keeps the
behavior consistent so that the relationships of a service must be empty
before it is destroyed.
If a listener was created at runtime and at some point it fails (e.g. the
port is already taken), the listener would be removed from the service. In
2.2, the removal of the listeners simply marked the listener as
inactive. In 2.3, the functions were combined so that the listener is
marked as inactive and removed from the workers. The fact that the
listener had a NULL DCB at that point caused the crash.
The correct thing to do is to not remove the listener from the service and
to mark the listener as inactive in the close_port helper function. The
listener will be freed once the service is destroyed.
If the filter does not declare defaults and no parameters are defined, the
list of parameters would be NULL. This was interpreted as an error and the
filter creation failed.
If the configuration processing encounters an error at the object
construction stage, it needs to stop immediately. If another object
depends on the object that failed, it would also fail but in a very
confusing manner. Mainly this manifests itself as a missing reference to
the object which would cause misleading errors to be logged.
