Summary:
Micro benchmarks show improvements in performance of MapLiteral from 5%
to 40% depending on the size of the input. On the other hand, a sequence
of AdditionOperators behaves the same with both allocation schemes.
Reviewers: mtomic, mferencevic, msantl
Reviewed By: mtomic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D2132
Summary:
The global variable may hide the fact that it uses the default
utils::NewDeleteResource() for allocations.
Reviewers: mtomic, llugovic, mferencevic
Reviewed By: mtomic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D2121
Summary:
`dump()` awesome Memgraph function is removed.
`mg_dump` executable now uses "dump database" command instead of "return dump()"
and dumps database to standard output instead of a file.
Reviewers: teon.banek
Reviewed By: teon.banek
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D2122
Summary:
Stream queries to the output table.
For effective output streaming, a simple operator, `OutputTableStream` is implemented which fetches and
produces a single row on each Pull.
Reviewers: teon.banek
Reviewed By: teon.banek
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D2099
Summary:
This diff only introduces a MemoryResource member to TypedValue and correctly
propagates through various constructors and assignments. At the moment,
MemoryResource is not used to actually allocate anything.
Reviewers: mtomic, llugovic, mferencevic, msantl
Reviewed By: mtomic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D2085
Summary:
HA should now support constraints in the same way the SM version does.
I only tested this thing manually, but I plan to add a new integration test for
this also.
Reviewers: ipaljak, vkasljevic, mferencevic
Reviewed By: ipaljak, mferencevic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D2083
Summary:
Implemented a simple dump executable which runs `RETURN dump()` on the database using Bolt protocol.
The `dump()` function returns a list of openCypher queries needed for database reconstruction (note: in the future we should use more appropriate method that supports streaming of data).
Reviewers: teon.banek, mferencevic
Reviewed By: teon.banek
Subscribers: pullbot, mferencevic
Differential Revision: https://phabricator.memgraph.io/D2067
Summary:
During it's leadership, one peer can receive RPC messages from other peers that his reign is over.
The problem is when this happens during a transaction commit.
This is handled in the following way.
If we're the current leader and we want to commit a transaction, we need to make sure the Raft Log is replicated before we can tell the client that the transaction is committed.
During that wait, we can only notice that the replication takes too long, and we report that with `LOG(WARNING)` messages.
If we change the Raft mode during the wait, our Raft implementation will internally commit this transaction, but won't be able to acquire the Raft lock because the `db.Reset` has been called.
This is why there is an manual lock acquire. If we pick up that the `db.Reset` has been called, we throw an `UnexpectedLeaderChangeException` exception to the client.
Another thing with long running transactions, if someone decides to kill a `memgraph_ha` instance during the commit, the transaction will have `abort` hint set. This will cause the `src/query/operator.cpp` to throw a `HintedAbortError`. We need to catch this during the shutdown, because the `memgraph_ha` isn't dead from the user perspective, and the transaction wasn't aborted because it took too long, but we can differentiate between those two.
Reviewers: mferencevic, ipaljak
Reviewed By: mferencevic, ipaljak
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1956
Summary:
Micro benchmarks show no change compared to global new & delete. This is
to be expected, because Unwind relies only on `std::vector` which ought
to reserve the memory in reasonable chunks.
Reviewers: mtomic, llugovic
Reviewed By: mtomic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D2064
Summary:
Micro benchmarks show an improvement to performance of about 10%
compared to global new & delete.
Reviewers: mtomic, llugovic
Reviewed By: mtomic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D2061
Summary:
Micro benchmarks show that MonotonicBufferResource improves performance
by a factor of 1.5.
Reviewers: mtomic, mferencevic, llugovic
Reviewed By: mtomic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D2048
Summary:
Benchmarks show minor improvements. Perhaps it makes sense at some later date
to use another allocator for things lasting only in a single `Pull`.
Reviewers: mferencevic, mtomic, llugovic
Reviewed By: mferencevic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D2018
Summary:
Unfortunately, the written micro benchmark only reports minor
improvements compared to default allocator. The results are in some
cases even a tiny bit worse.
Reviewers: mtomic, mferencevic, llugovic
Reviewed By: mtomic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D2039
Summary:
# Summary
## Concepts and terminology
A **namestring for <object>** is a Lisp string that names the C++ language element
`<object>` (such as a namespace, a variable, a class, etc.). Therefore we have a
"namestring for a namespace", a "namestring for a variable", etc.
A **designator for a namestring for <object>** is a Lisp object that denotes a
"namestring for <object>". These are symbols and strings themselves.
A **typestring** is a Lisp string that represents a C++ type, i.e. its
declaration.
A **typestring designator** is a Lisp object that denotes a typestring.
A typestring (and the corresponding C++ type) is said to be **supported** if it
can be parsed using `parse-cpp-type-declaration`. This concept is important and
should form the base of our design because we can't really hope to ever support
all of C++'s type declarations.
A typestring (and the corresponding C++ type) that is not supported is
**unsupported**.
A **processed typestring** is a typestring that is either fully qualified or
unqualified.
A C++ type is said to be **known** if LCP knows extra information about the type,
rather than just what kind of type it is, in which namespace it lives, etc. For
now, the only known types are those that are defined within LCP itself using
`define-class` & co.
A C++ type is **unknown** if it is not known.
**Typestring resolution** is the process of resolving a (processed) typestring
into an instance of `cpp-type` or `unsupported-cpp-type`.
**Resolving accessors** are accessors which perform typestring resolution.
## Changes
Explicitly introduce the concept of supported and known types. `cpp-type` models
supported types while `unsupported-cpp-type` models unsupported types.
Subclasses of `cpp-type` model known types. `general-cpp-type` is either a
`cpp-type` or an `unsupported-cpp-type`.
Add various type queries.
Fix `define-rpc`'s `:initarg` (remove it from the `cpp-member` struct).
Introduce namestrings and their designators in `names.lisp`.
Introduce typestrings and their designators.
Introduce **processed typestrings**. Our DSL's macros (`define-class` & co.)
convert all of the given typestrings into processed typestrings because we don't
attempt to support partially qualified names and relative name lookup yet. A
warning is signalled when a partially qualified name is treated as a fully
qualified name.
The slots of `cpp-type`, `cpp-class`, `cpp-member` and `cpp-capnp-opts` now
store processed typestrings which are lazily resolved into their corresponding
C++ types.
The only thing that instances of `unsupported-cpp-type` are good for is getting
the typestring that was used to construct them. Most of LCP's functions only
work with known C++ types, i.e. `cpp-type` instances. The only function so far
that works for both of them is `cpp-type-decl`.
Since "unsupportedness" is now explicitly part of LCP, client code is expected
to manually check whether a returned type is unsupported or not (or face
receiving an error otherwise), unless a function is documented to return only
`cpp-type` instances.
A similar thing goes for "knowness". Client code is expected to manually check
whether a returned type is known or not, unless a function is documented to
return only (instances of `cpp-type` subclasses) known types.
## TODO
Resolution still has to be done for the following slots of the
following structures:
- `slk-opts`
- `save-args`
- `load-args`
- `clone-opts`
- `args`
- `return-type`
Reviewers: teon.banek, mtomic
Reviewed By: teon.banek
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1962
Summary:
According to the written benchmark, using MonotonicBufferResource yields
significant improvements to performance of Distinct. The setup fills the
database with vertices depending on the benchmark state. No edges are
created. Then we run DISTINCT on that. Since each vertex is unique, we
will store everything in the `DistinctCursor::seen_rows_`, which is
backed by a MemoryResource. This setup, on my machine, yields 10 times
better performance when run with MonotonicBufferResource.
Reviewers: mferencevic, mtomic, msantl
Reviewed By: mferencevic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1894
Summary:
There will be a lot of leftover files, execute the following commands inside
`src/` to remove them:
```
git clean -xf
rm -r rpc/ storage/single_node_ha/rpc/
```
Reviewers: teon.banek
Reviewed By: teon.banek
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D2011
Summary:
This API change is needed in order to propagate the memory allocation
scheme for the execution of LogicalOperator::Cursor
Depends on D1990
Reviewers: mtomic, mferencevic
Reviewed By: mtomic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1980
Summary:
The new distributed directory is inside the query, and mirrors the query
structure. This groups all of the distributed (query) source code
together, which should make the potential directory extraction easier.
Reviewers: mferencevic, llugovic, mtomic
Reviewed By: mferencevic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1923
Summary:
This is a bugfix for D1836. It made `SymbolTable` return references to vector
elements, which then get invalidated and weird stuff happens.
This made a `DCHECK` in `rule_based_planner.hpp` trigger, and it was noticed by
@ipaljak 2 months later. All `DCHECK`s in `rule_based_planner.hpp` are now
changed to `CHECK`s.
Also, hash function for `Symbol` was wrong, because it also took
`user_declared` field into consideration, and `==` operator doesn't do that.
Reviewers: ipaljak, teon.banek, mferencevic, msantl
Reviewed By: msantl
Subscribers: pullbot, ipaljak
Differential Revision: https://phabricator.memgraph.io/D1938
Summary:
`SHOW STORAGE STATS` when executed in a Raft cluster should return
stats for each member of the cluster.
`StorageStats` starts a RPC server on each member of the cluster that answers
about its local storage stats.
The query can be invoked only on the current leader, the leader sends a request
to each peer and shows the results it gets. If some peers don't answer within 1
second, stats for those peers won't be shown.
The new output can be seen here: P27
Reviewers: ipaljak, mferencevic
Reviewed By: ipaljak
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1907
Summary: Removing Kafka, auth and audit log features from HA instance for the time being.
Reviewers: mferencevic, msantl
Reviewed By: mferencevic, msantl
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1882
Summary:
`EdgesIterable` is used for iterating over Edges in distributed memgraph. Because of lru cache there is a possibility of data getting evicted as someone iterates over it.
To prevent that `EdgesIterable` will lock that data and release it when it's deconstructed.
Reviewers: msantl
Reviewed By: msantl
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1868
Summary:
Added a new awesome function called `storageInfo`. This function
returns a list of key-value pairs containing the following (useful?)
information:
* number of vertices
* number of edges
* average degree
* memory usage
* disk usage
The current implementation is in `awesome_memgraph_functions` but it will end up
as a separate clause for better access control.
Reviewers: teon.banek, mtomic, mferencevic
Reviewed By: teon.banek
Subscribers: pullbot, buda
Differential Revision: https://phabricator.memgraph.io/D1850
Summary:
RuleBasedPlanner now generates only the regular ScanAll operations, and
Filter operations are appended as soon as possible. The newly added
Rewrite step, takes this operator tree and replaces viable Filter &
ScanAll operators with appropriate ScanAllBy<Index> operator. This
change ought to simplify the behaviour of DistributedPlanner when that
stage is moved before the indexed lookup rewrite.
Showing unoptimized plan in interactive planner is also supported.
Reviewers: mtomic, llugovic
Reviewed By: mtomic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1839
Summary:
All AST nodes had a member `uid_` that was used as a key in
`SymbolTable`. It is renamed to `symbol_pos_` and it appears only in
`Identifier`, `NamedExpression` and `Aggregation`, since only those types were
used in `SymbolTable`. SymbolGenerator is now responsible for creating symbols
in `SymbolTable` and assigning positions to AST nodes.
Cloning and serialization code is now simpler since there is no need to track
UIDs.
Reviewers: teon.banek
Reviewed By: teon.banek
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1836
Summary:
Variable expansions cannot appear in merge patterns or after updates,
so they can only be planned with GraphState::OLD. Because of that, it makes
sense to remove GraphView parameter from them to reduce confusion.
Reviewers: teon.banek, llugovic
Reviewed By: teon.banek
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1825
Summary:
Implement proper plan cloning using LCP instead of hacking it with
serialization.
depends on D1815
Reviewers: teon.banek, llugovic
Reviewed By: teon.banek
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1816
Summary:
use newly added LCP functionality to get rid of manually written
`Clone` functions in AST.
depends on D1808
Reviewers: teon.banek, llugovic
Reviewed By: teon.banek
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1815
Summary:
When there was an empty line comment starting right at the end of the
query, stripper wouldn't properly change state from `IN_LINE_COMMENT` to `OUT`
and it would return the wrong length (0) from `MatchWhitespaceAndComments`.
Because of that, the two slashes would be interpreted as division operators.
Query "RETURN 5;//" would be changed by stripper into "RETURN 5; / /" which
obviously can't be parsed;
Reviewers: teon.banek
Reviewed By: teon.banek
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1819
Summary:
Rename Context to ExecutionContext and make it struct
Move ParsingContext to cypher_main_visitor.hpp
Reviewers: mtomic, llugovic
Reviewed By: llugovic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1810
Summary:
Identifiers would be printed correctly if they were printed as part of a larger
`Expression` (so the visitor did the printing). However, if `PrintObject` was
called with an `Identifier *` then it would delegate to the template overload
and just print the pointer.
Reviewers: mtomic, teon.banek
Reviewed By: mtomic
Subscribers: mferencevic, pullbot
Differential Revision: https://phabricator.memgraph.io/D1802
Summary:
This diff removes the need for a database when parsing a query and
creating an Ast. Instead of storing storage::{Label,Property,EdgeType}
in Ast nodes, we store the name and an index into all of the names. This
allows for easy creation of a map from {Label,Property,EdgeType} index
into the concrete storage type. Obviously, this comes with a performance
penalty during execution, but it should be minor. The upside is that the
query/frontend minimally depends on storage (PropertyValue), which makes
writing tests easier as well as running them a lot faster (there is no
database setup). This is most noticeable in the ast_serialization test
which took a long time due to start up of a distributed database.
Reviewers: mtomic, llugovic
Reviewed By: mtomic
Subscribers: mferencevic, pullbot
Differential Revision: https://phabricator.memgraph.io/D1774
Summary:
Before I wrongly assumed `Shutdown` will always be called on Cursors and
removed BFS subcursors there. Now it is done using transactional cache clean-up
mechanism. There's a separate clean-up for `BfsSubcursorStorage` (for actual
subcursors) and `BfsRpcServer` (for database accessors created by the server).
I've changed `BfsRpcServer` to have a `GraphDbAccessor` per transaction,
instead of per `BfsSubcursor`. Mainly because there is no reliable way to check
if the transaction tied to the accessor has expired as it is not safe to call
`transaction_id` method (since `GraphDbAccessor` is holding only a reference to
`Transaction` object).
Reviewers: teon.banek, mferencevic
Reviewed By: teon.banek
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1792
Summary:
Created a new integration test for Raft protocol.
The tests iterates through the Raft cluster and does the following:
* kill machine `X`
* execute a query
* bring `X` back to life
The first step is to insert a vertex in the cluster, and last step is to check
if the cluster has all the data.
I also edited some of the raft core files because this test surafaced some bugs.
The `tester` binary is a hacked version of the HA client and so are the parts in
the code that refuse to execute a query is the machine is not in `Leader` mode.o
Those parts will go away once we have a proper HA client.
I've run the `runner.py` for a while (215 times)
```
while ./runner.py &> log.txt; do echo -n "."; done
```
and it didn't break.
Reviewers: ipaljak, mferencevic
Reviewed By: ipaljak
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1788
Summary: This simple function is required by the Tensorflow integration so that Memgraph can always return regular matrices of desired size.
Reviewers: teon.banek, mtomic, dsantl
Reviewed By: mtomic, dsantl
Subscribers: mferencevic, pullbot
Differential Revision: https://phabricator.memgraph.io/D1783
Summary:
This (almost) removes the dependency of operators on NodeAtom and
EdgeAtom. Only EdgeAtom::Direction is needed. The change was done as the
initial step of removing dependency on storage from Ast. Additionally,
it makes sense for LogicalOperator to only depend on Expression classes.
Reviewers: mtomic, llugovic
Reviewed By: llugovic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1779
Summary:
Classes marked with `:serialize (:slk)` will now generate SLK
serialization code. This diff also changes how the `:serialize` option
is parsed, so that multiple different serialization backends are
supported.
Reviewers: mtomic, llugovic, mferencevic
Reviewed By: mtomic
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D1755