* make replication timeouts configurable via startup options
The following options are available (for active failover
and master-slave replication):
--replication.connect-timeout
--replication.request-timeout
Values can be specified in seconds. If these options are used, they will
be used for replication requests, overriding any hard-coded defaults or
explicitly configured timeouts.
Additionally, this change increases the default request timeout
for replication from 10 minutes to 20 minutes.
* do *not* change timeouts
* make tests work again
* Fixed repeatable seek to the same document (#10164)
* Fixed repeatable seek to the same document
* Fix. Multiple conjuncts in conjunction.
* Add tests for issue #10090 (#10203)
* Add tests for issue #10090
* Remove spaces
* Added CHANGELOG for the issue #10090.
* Fixed doc_limits error
* Fixed approx_ error
* Feature/rebootid notice changes, backport of #9523
* Move to 3.4 and C++11-compatible code (except for test code)
* Backported tests/Mocks/Servers.{h,cpp}
* Rebuilt errorfiles
* Ported CallbackGuardTest from gtest to catch
* Ported RebootTrackerTest from gtest to catch
* Make sure the state method is called, so the overridden method is used during tests
* Fixed test to work with the old scheduler
* release version 3.4.8
* [3.4] fix agency lockup when removing 404-ed callbacks (#9839)
* Update arangod/Cluster/ServerState.cpp
Co-Authored-By: Markus Pfeiffer <markuspf@users.noreply.github.com>
* Instantiate the scheduler during ::prepare()
* Fix test crash introduced during backport
* Fix a compile error on Windows (thanks, V8)
* Bug fix/add shard id to replication client identifier (#9366)
* Fixed compile (but not linker) errors
* Backported ReplicationClientProgressTracker
* Fixed compile errors, fixed hash function
* No longer use SyncerId for real asynchronous replication
* Updated docu
* Try to fix compile error on windows
* Fixed a bug
* Removed old code
* Fixed CHANGELOG entry
* op:set should have new, some operators should not create non existing keys
* this is the proper direction
* observe/unobserve have no business in Node
* normalization of URI
* agency test added for observe / unobserve
* more tests
* Fixed catch tests.
* Added full index iterator utilization in case of some documents was not found in collection
* Added index operations reversal in case of insertion failure in rocksdb
* Removed singleop optimization as rollback possible even for such operations (in case of index operation failure on insert)
* Added tests for insert rollback in views
* Added catch tests for view block
* Added changelog entry
* fixed jslint errors
* fixed linux build
* Added test for update
* Fix CHANGELOG sequence
* Add line
* Added reverts for update and remove failures
* Cleanup in test. Added throwing error in case of failed reversal
* fixed typo
Co-Authored-By: Simran <Simran-B@users.noreply.github.com>
* fixed build
* Applied review comments
* Applied review comments
* preview code I dropped on jackie today. only supports document collections.
* saving work in progress
* update arangoexport to fully use ManagedDirectory object (arangoimport still needs work).
* migrate remainder of import to using ManagedDirectory objects
* correct spelling within activateEncryption(): encription to encryption
* forgot to fix up graphExport for managed directory. luckily unittest noticed.
* add test read of gzip json file
* add readGzip to our fs object
* add TRI_SlurpGzipFile function
* add code to bridge javascript fs.readGzip to TRI_SlurpGzipFile
* correct the return type used by JS_ReadGzip
* add gzip test for jsonl type export
* add remain gzip versions of existing export tests
* add import tests of gzip files
* activate encryption in enterprise to enable fs.readDecrypt()
* bug fix: allow encryption to enable in empty directory if EncryptionFeature active.
* add JS_ReadDecrypt to enterprise build. Current used for unit tests.
* enterprise build needs EncryptionFeature added to build to support TRI_SlurpDecryptFile()
* add TRI_SlurpDecryptFile() to aid in unit test of encrypted stuff
* missing newline at end of file.
* add fs.readDecrypt when available
* add test of encrypted json array export
* correct progress meter when gzip involved
* move from using ManagedDirectory::File to ManagedDirectory::readableFile calls
* create version of readableFile that uses file descriptor as param
* Cleanup new logging.
* Hand on error message from getLocalCollections.
* Better behaviour if a database was announce but has vanished since.
* Fix catch tests.
* Switch on maintenance debugging output.
* Fix maintenance reporting bugs.
* CHANGELOG.
* No error if follower cannot be dropped.
* Improvement to avoid copying.
* Add preconditions to FollowerInfo agency operations.
* Adjust timeouts.
* Use isEqualString instead of compareString.
* Fix Windows compilation.
* Increased timeouts, added a little more helpful error messages for assertions (#9135)
* Wait for replication before inserting documents
Un-greylist repair-distribute-shards-like-spec
* fix flientid register in state
* remove the log erase locations
* iterator check
* also clear considered
* emplaces also handled
* don't add if clientId empty
* move semantics for log_t
* forward log_t into deque. bail out also as follower, when log_t cannot be emplaced
* whitespaces
* change log
* fix agency test for leadership change during long transactions with clientId
* log_t was as good as it gets
* white spaces
* add clientIds the inc/dec test
* client id for inc/dec test
* Fix a typo in a message.
* Add a comment for better readability.
* Fix agency test.
* Take debugging out.
* Activate another old agency test.
* Fix client id.
* Consistent behaviour across branches.
Jan
Iurii Popov
Tobias Gödderz
KVS85
Frank Celler
Heiko
Dronplane
Dan Larkin-York
Lars Maier
Max Neunhöffer
Kaveh Vahedipour
Jan Christoph Uhde
Wilfried Goesgens
Matthew Von-Maszewski
jsteemann
Tobias Gödderz
Simon