Added the `transaction_replay_max_size` parameter that controls the
maximum size of a transaction that can be replayed. If the limit is
exceeded, the stored statements are released thus preventing the
transaction from being replayed.
This limitation prevents accidental misuse of the transaction replaying
system when autocommit is disabled. It also allows the user to control the
amount of memory that MaxScale will use.
As the current query was added to the transaction log before it finished,
the m_current_query contained a duplicate of the latest transaction log
entry. To correctly log only successful transactions, the statement should
be added only after it has successfully completed. This change also
removed the unnecessary cloning that took place when the statement was
added to the log before it finished.
With the fixed transaction logging, the value of m_current_query can be
stashed for later retrying while the replay process is happening. If the
replay completes successfully and the checksums match, the interrupted
query is retried.
Also added a clarifying comment to can_retry_query to explain why a query
inside a transaction cannot be retried.
Added the initial implementation of transaction replay. Transactions are
only replayed if the master fails when no statement is being executed.
The validity of the replayed transaction is done by verifying that the
checksums of the returned results are equal.
Added a close function into the Trx class to make resetting its state
easier. Also changed the return type of the pop_stmt to GWBUF* as the
places where it is used expect a raw GWBUF pointer.
The queries that make up the transaction are now stored in the router
session while the transaction is in progress. For the time being, the
queries are only used to log extra information about the transaction
contents.
Readwritesplit now calculates checksums for all successful and failed
transactions. This checksum is not of any practical use until the
transaction replaying is implemented.
Only servers that qualify to be connected should be considered as
candidate servers. This triggered a debug assertion when a slave server
failed to execute a session command but it was chosen as a candidate
server later on.
The state could be factored out into a boolean variable as the reply
processing can be in two states: Either waiting for the response to
MASTER_GTID_WAIT or updating packet numbers.
The packet number updating can always be done as long as a buffer is
available. The discard_master_wait_gtid_result function discards the OK
packet before the packet numbers are updated so any trailing packets get
corrected properly.
If a query is interrupted that was sent to the master, it is now
retried. This allows all autocommit queries to be transparently retried if
the server in question fails.
By adding a boolean parameter, the feature can be enabled with sensible
default values. Renamed the query_retry parameters and set the defaults to
acceptable values.
Added new parameters to diagnostic output. Also did some minor renaming.
This is a proof-of-concept that validates the query retrying method. The
actual implementation of the query retrying mechanism needs more thought
as using the housekeeper is not very efficient.
Keeping track of the closed state of the session inside the router session
itself is not needed as the MaxScale core should already do that.
The skygw_chk_t variables are rather meaningless and are obsoleted by
Valgrind/ASAN.
Most of the funtionality is now a member function of either the RWSplit or
RWSplitSession class. This removes the need to pass the router and session
parameters to all functions.
Moved the RWBackend class implementation into its own file. Made some of
the command type functions a part of the <maxscale/protocol/mysql.h>
header to make it reusable.
The `master_reconnection` parameter now controls both the reconnection of
the master server as well as the migration of the master server to another
server. Although these two cases appear to be different, the end result
from readwritesplit's point of view is the same and are thus controlled
with the same parameter.
The RWBackend class now resets its internal state when it is closed. This
allows readwritesplit to handle the case when a result was expected from
the master but the master died before the result was returned. The same
code should also handle slave connection failures mid-result, allowing
Backend reuse.
Added a test case that verifies the new functionality when combined with
`master_failure_mode=error_on_write`.
Provides a clearer separation between what deals with query
classification and what deals with query routing.
Functions have only been moved. No other cleanup has been
done.
The responses of slaves that arrived before the master were always
compared to the empty value of 0x00. If the slave connection replied after
the master, the comparison was correct.
This commit introduces a map of slaves and their responses that
are handled once the master's response arrives.
* MXS-199: Support Causal Read in Read Write Splitting
* move most causal read logic into rwsplit router and get server type from monitor
* misc fix: remove new line
* refactor, move config to right place, replace ltrim with gwbuf_consume
* refacter a little for previous commit
* fix code style
The result collection did not reset properly when a non-resultset was
returned for a request. As collected result need to be distinguishable
from single packet responses, a new buffer type was added.
The new buffer type is used by readwritesplit which uses result collection
for preparation of prepared statements.
Moved the current command tracking to the RWBackend class as the command
tracked by the protocol is can change before a response to the executed
command is received.
Removed a false debug assertion in the mxs_mysql_extract_ps_response
function that was triggered when a very large prepared statement response
was processed in multiple parts.
Inlined the getter/setter type functions that are often used. Profiling
shows that inlining the RWBackend get/set functions for the reply state
manipulation reduces the relative cost of the function to acceptable
levels. Inlining the Backend state function did not have as large an
effect but it appears contribute a slight performance boost.
The result processing code did unnecessary work to confirm that the result
buffers are contiguous. The code also assumed that multiple packets can be
routed at the same time when in fact only one contiguous result packet is
returned at a time.
By assuming that the buffers are contiguous and contain only one packet,
most of the copying and buffer manipulation can be avoided.
