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.
The functions
bool runtime_is_string_or_null(json_t* json, const char* path);
bool runtime_is_bool_or_null(json_t* json, const char* path);
bool runtime_is_count_or_null(json_t* json, const char* path);
can now be called from anywhere inside MaxScale.
In principle it would be better if the qc information were
obtained via a specific query_classifier resource. However,
there are multiple problems with that (e.g. the qc has no way
of safely accessing information of another thread) and hence
the worker specific qc cache statistics is reported as part of
the worker statistics.
The service capabilities were not set for filters as the name overlapped
with a local variable. The variable name was fixed but the actual use of
it was not.
If a passwd parameter is defined in a persisted configuration, it will be
ignored. This will make upgrades from 2.2 to 2.3 possible with persisted
configurations.
If a relationship isn't defined, it should not be removed. Only if a
relationship is change to a null relationship, should it be removed.
Also fixed the maxctrl test suite to verify that both it and the REST API
tests pass.
The data of the relationship should be set to null to delete the
relationship. Conceptually it is different from setting it to an empty
relationship but in practice both lead to the same end result.
Updated tests to use correct relationships and added test cases for
detection of invalid relationships.
The testing framework extended the public struct, not the private
one. Also moved the internal Session class inside the mxs namespace to
prevent conflicts with the mock testing Session class.
The filter relationships of a service are now interpreted as an ordered
list of filters being used by the service. As the relationship data in the
JSON API specification is represented as an array, it does not prevent
multiple relationships from appearing. For MaxScale, this means that
interpreting the filter's order does not break it.
As the filters are only passed as a pipe separated list when the
configuration is being processed, there's no need to have the interface
conform to that. Passing a list of filter names makes it more flexible and
will make it's use in the runtime configuration easier.
The Session class now contains all of the C++ objects that were previously
in the MXS_SESSION struct. It is also allocated with new but all
initialization is still done outside of the Session in session_alloc_body.
This commit will not compile as it is a part of a set of commits that make
parts of the session private.
The service now generates the configuration file that represents the
serialized form of it. Also removed the unused service_serialize_servers
function.
The services will now store a local copy of the filter lists in the worker
local data associated with the service. This removes the instance level
lock and removes the performance penalty that was previously imposed by
it.
The only remaining performance "regression" compared to 2.2 is the extra
two atomic operations per filter that a session does when it is being
created. This is quite hard to get rid of without significant amounts code
and will hopefully be just a drop in the ocean.
The most relevant string variables of a service are now duplicated as C++
strings. This should ease the eventual transition to a fully C++ internal
representation of the service. The array of refresh rates was also wrapped
inside a std::vector to remove the need to manually manage memory.
Separated the SERVICE_USER struct into its individual components as there
was no real need to have them inside a struct.
The filters can now be altered at runtime. Currently, the filter usage
uses a service level lock. The next is to allocate a per-service key to
the worker local data and use that to orchestrate the storage of the
filter lists.
Removed the explicit setters for the service parameters. Not all of them
were implemented and they were only used internally. Moved the parameter
validation and update processing inside the Service class to reduce the
load on the other parts of the core.
The service now has a private std::mutex that is used for
synchronization.
Renamed the vector of services to use snake_case.
Use lock guards with mutexes to make usage easier and safer. This makes
the code smaller as well as slightly easier to read.
The service now uses a std::vector<SFilterDef> to store the filters it
uses. Most internal parts deal with the SFilterDef but debugcmd.cc still
moves raw pointers around (needs to be changed).
The FilterDef structs are now stored in a vector<std::shared_ptr>. This
should make it easier to use filters even if they are deleted before the
session using them closes.
All internal functions now take a smart point as a parameter. One
problematic case will be debugcmd.cc which moves information around as
pointers cast into unsigned longs.
