Summary:
There were a couple of issues with handling the above 2 signals.
1) Calling `std::exit` from a signal handler is undefined behaviour.
The only defined way for a signal handler to stop the program is calling
one of: `std::_Exit`, `std::abort` or `std::quick_exit`. Neither of them
will completely clean the resources, so a clean exit is not possible.
Since SIGSEGV and SIGABRT happen in extraordinary circumstances that we
wish to debug 99% of the time, it makes sense to generate a core dump
which can be inspected by a debugger. Of the 3 termination functions,
only `std::abort` will generate a core dump, so it makes sense to use
that to stop the program.
Also, since we are now aborting as is the default behaviour on SIGSEGV
and SIGABRT, it becomes questionable why have a custom handler at all.
2) Raising an exception inside a signal handler is undefined behaviour
Although the handler by itself does not raise an exception, it is
possible for the logging facility to raise one. This is a real case when
logging a stack trace in particular. Stack trace is generated by
creating a string "<function name> <line location>". It is possible that
a function name will contain '{}' somewhere inside. This is usually the
case with anonymous functions. The generated string is then passed to
logging, which uses the `fmt` library to fill '{}' with remaining
arguments. Since only a single argument (the stack trace string) is
passed for formatting, naturally the `fmt::format` throws an exception,
that it is missing a format argument.
We could provide an overload which takes a single string, but that
defeats the purpose of `fmt::format` raising an exception in regular
code if we forget to pass an argument. Another solution is to escape the
whole stack trace string, so it is valid for formatting, but this only
complicates the handler even further. The simplest solution is to send
the stack trace to `stderr` and avoid logging altogether.
Simplify Shutdown, so it can be used in a signal handler
Reviewers: florijan, mferencevic, buda, mislav.bradac
Reviewed By: mferencevic, buda
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D474
Summary:
Remove utils/stacktrace/log.hpp
The single function defined in log.hpp is used only in
memgraph_bolt.cpp. Therefore, the function has been moved and the file
removed.
Move utils/stacktrace/stacktrace.hpp one level up
Move some logging from memgraph_bolt to Server
Reviewers: buda, dtomicevic
Reviewed By: buda
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D465
Summary:
The rest of core engine exceptions are extended from the BasicException (because of the error handling in the communication stack). Maximum value of the MVCC command id is calculated in the compile time.
I've put the tasks to implement concurrent tests in the qa repo in the backlog. Hopefully they will be implemented within this sprint.
Reviewers: mislav.bradac, mferencevic, florijan
Reviewed By: mislav.bradac
Subscribers: dgleich, pullbot, buda
Differential Revision: https://phabricator.memgraph.io/D298
Summary:
Bolt buffer is now a template.
Communication worker now has a new interface.
Fixed network tests to use new interface.
Fixed bolt tests to use new interface.
Added more functions to bolt decoder.
Reviewers: dgleich, buda
Reviewed By: buda
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D256
Summary:
This should fix the integration query engine test. I'm not exactly sure what caused the SEGFAULT to occur, but from my debugging sessions; I would say, as an educated guess that it's closely related to logger. What I believe happens is that logger gets somehow initialized from the dynamic_library when it's opened. And when the library is closed the corresponding logger is freed, but, other parts of memgraph still see that instance of logger and trying to use to emit code makes the whole thing crash. I've removed one include of logger from something which gets included in the hardcoded queries, but I noticed there are some more.
I've changed the dlopen to now open with DEEPBIND, which should cause symbols to be resolved locally and not globally and stop the logger from the dynamic lib to interfer with the one used in the rest of memgraph.
Reviewers: buda
Reviewed By: buda
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D234
Summary: Bolt PullAll works. ChunkedBuffer is implemented by the specification (the tests can be found in the tests/unit/bolt_chunked_buffer.cpp). Bolt session states method are refactored (all run methods have only one argument - active Session). Modifications related to the style guide.
Reviewers: mferencevic
Reviewed By: mferencevic
Subscribers: pullbot, buda
Differential Revision: https://phabricator.memgraph.io/D180
Summary:
Removed old encoder.
Changed namespace from bolt to communication::bolt.
Removed old include from new encoder.
Added an empty message success to encoder.
Changed order in communication::Server.
Changed bolt session to use new encoder.
Merge remote-tracking branch 'origin/dev' into mg_hardcoded_queries
Fixed PrintRecordStream.
Reviewers: buda, dgleich
Reviewed By: buda
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D158
Summary:
Moved server and worker from bolt to communication. Started
templatization.
Started removal of Bolt class.
Removed unnecessary files from network.
Converted states to template functions.
Bolt::Session is now a template.
Merge remote-tracking branch 'origin/dev' into mg_refactor_network
Merged bolt_serializer.cpp into hpp.
Removed obsolete include.
Initial version of bolt session unit test.
Uncommented leftover log commands.
Reimplemented io::Socket.
Added client-stress.sh script.
Reviewers: dgleich, buda
Reviewed By: dgleich, buda
Subscribers: pullbot, mferencevic, buda
Differential Revision: https://phabricator.memgraph.io/D64
Summary: database/GraphDbAccessor and storage/...Accessor now returns iterators over lazily created Accessors (filtered on visibility). Dependecies and forward declarations reorganized to support new code
Reviewers: buda, teon.banek
Reviewed By: buda
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D83
Summary:
Rename TypedValue to PropertyValue.
Move original TypedValue to query/backend/cpp.
Fix undefined behaviours and memory leaks in PropertyValue.
Add list type to PropertyValue.
Operators and appropriate tests will be revised in following commits.
Fix bugs in TypedValue
Fix bugs in typed value
Reviewers: buda, florijan
Reviewed By: buda
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D77
Summary: For each hardcoded query plan there is a build target -> that will improve Memgraph's testability + update of release script (release/alpha.sh)
Reviewers: florijan, mislav.bradac, mferencevic, dgleich
Reviewed By: dgleich
Subscribers: pullbot, buda
Differential Revision: https://phabricator.memgraph.io/D75
Summary:
Cast pack::Code enum to byte and size_t where neccessary. g++ is
confused by some of implicit casts when function has different
valid overloads. There is still problem in states/init.cpp but that file
will disappear after bolt is refactored. Next step towards safe code is
to make Code and Rule enum class.
Reviewers: buda, mferencevic
Reviewed By: buda
Subscribers: pullbot, mferencevic, buda
Differential Revision: https://phabricator.memgraph.io/D58
Summary:
Begin record stream mocker implementation.
Record stream mocker alongside tests created.
Test Plan: Run record_stream_mocker in tests.
Reviewers: mislav.bradac, florijan, buda
Reviewed By: buda
Subscribers: pullbot
Differential Revision: https://phabricator.memgraph.io/D54
tmp commit
tmp commit v2
Finished reimplementation of propertys.
They now can be placed in a holder with different source of type information.
Tmp commit
