CheckExtraDataStores reboots or kills storage servers with extra data stores.
Since this occurs during a consistency check, the expectation is that the database
is quiet and not in the midst of recovery. This was done with a 10-second delay,
but it's possible during simulation tests that it takes longer than 10 seconds
to recruit a new master, so this assumption is invalid and can cause a test failure
when the consistency checks proceed.
Instead of a delay, we run an empty transaction through the system and explicitly
wait for the cluster to return to a fully-recovered state.
When VV is enabled, the waitForVersion() may use the oldest version in the MVCC window as the actual read version, which may often result in the transaction_too_old if the MVCC moved, especially during a readRange(). The change here is to try using a higher read version that is still valid, but not likely to cause transaction_too_old error.
* fix: the log router could see popped() versions between recoveredAt and the recovery txn version which cause remote tlogs to be stopped at a version in this range, which causes storage servers to read at versions in this range
* fix: a spurious correctness bug resulted from the consistency check running during a recovery
* Recruit new singleton for consistency checker.
* Recruit the consistency checker only if enabled.
* Add a yield in monitorConsistencyChecker().
* Minor fixes.
* Consistency check workload enhancements.
* Minor fixes and clarifications.
* clang format
* Clang format.
* Minor fixes, cleanup, debug tracing.
* Misc.
* Move the consistency scan information from dbconfig to a key backed object.
* Move consistency scan config out of db cofig to a state object and feature rename.
* ConsistencyCheck workload refactor.
* devFormat
* Update fdbcli/ConsistencyScanCommand.actor.cpp
* Review Comments.
Co-authored-by: negoyal <neelam.goyal@gmail.com>
Co-authored-by: Ata E Husain Bohra <ata.husain@snowflake.com>
* proof of concept
* use code-probe instead of test
* code probe working on gcc
* code probe implemented
* renamed TestProbe to CodeProbe
* fixed refactoring typo
* support filtered output
* print probes at end of simulation
* fix missed probes print
* fix deduplication
* Fix refactoring issues
* revert bad refactor
* make sure file paths are relative
* fix more wrong refactor changes
Previously to get around the issue that EKP is not present when restart test switching encryption from on to off and read encrypted data, EKP was made to start in simulation regardless of encryption knob. This PR revert that change, and instead force restart test not to change encryption knob, by passing previous encryption knob through restartInfo.ini file. Also since we don't allow downgrading an encrypted cluster to previous version, disable encryption in downgrade tests.
Also adding an assert to allow reading encrypted mutations only if encryption knob is on. We may reconsider allowing switching encryption on/off for existing cluster, but for now we don't allow it.
* Shard based move.
* Clean up.
* Clear results on retry in getInitialDataDistribution.
* Remove assertion on SHARD_ENCODE_LOCATION_METADATA for compatibility.
* Resolved comments.
Co-authored-by: He Liu <heliu@apple.com>
Adding GetEncryptCipherKeys and GetLatestCipherKeys helper actors, which encapsulate cipher key fetch logic: getting cipher keys from local BlobCipherKeyCache, and on cache miss fetch from EKP (encrypt key proxy). These helper actors also handles the case if EKP get shutdown in the middle, they listen on ServerDBInfo to wait for new EKP start and send new request there instead.
The PR also have other misc changes:
* EKP is by default started in simulation regardless of. ENABLE_ENCRYPTION knob, so that in restart tests, if ENABLE_ENCRYPTION is switch from on to off after restart, encrypted data will still be able to be read.
* API tweaks for BlobCipher
* Adding a ENABLE_TLOG_ENCRYPTION knob which will be used in later PRs. The knob should normally be consistent with ENABLE_ENCRYPTION knob, but could be used to disable TLog encryption alone.
This PR is split out from #6942.
Xiaoxi Wang