Add START_MOVE_KEYS_MAX_RETRIES knob (default 50, BUGGIFY=10) and throw
start_move_keys_too_many_retries when exceeded. Wire the new error into
DDRelocationQueue (same light-weight re-queue handler as
finish_move_keys_too_many_retries and move_to_removed_server) and
normalDDQueueErrors so DD does not restart. Also propagate actor_cancelled
before the retry limit check in startMoveKeys.
* DD 7.3 finish movekeys backoff (#12991)
* Add jitter to finishMoveKeys backoff
Jitter the exponential backoff delay to [0.75x, 1.25x] of the base
value. This prevents all 15 FlowLock slots from retrying in lockstep
when they all hit transaction_too_old at the same time.
* Cap finishMoveKeys backoff at 5s after jitter
Apply the 5.0s cap after jitter so the final delay never exceeds
the documented maximum.
Problem: some of these transactions have been observed to involve cascades of many shards trying to execute them simultaneously owing to somewhat unpredictable DD dynamics. This then results in DD being pegged on CPU with many failing transactions.
Solution: the counters here will make it clear if this is happening and if so, where, so that further mitigations/solutions can be devised. Some of these code paths have TraceEvents, but some don't and of the existing TraceEvents, many are sampled. This PR avoids all of that and counts every time through, making us less reliant on guesswork.
An alternative approach (not taken here for many reasons) would be to instrument transaction code to let callers pass a tag. Then have the transaction client and/or server side emit a few metrics (started, committed, aborted) parameterized by tag.
One may reasonably object that the boilerplate here is kind of fugly. I guess my view is that at this late date the time for more subtle approaches has come and gone. We need this code to tell us what it is doing and if this looks a little intrusive, so be it.
Testing:
20260422-234326-gglass-2f514436068d99e6 compressed=True data_size=35519571 duration=5298275 ended=100000 fail=1 fail_fast=10 max_runs=100000 pass=99999 priority=100 remaining=0 runtime=0:55:44 sanity=False started=100000 stopped=20260423-003910 submitted=20260422-234326 timeout=5400 username=gglass
* Add DD init visibility, metrics retries, shard tracking, scan progress, and team collection logging (#12913)
Add logging throughout DD startup and runtime to diagnose slow startups,
stuck data moves, and undesired server classification.
DDTxnProcessor: Log elapsed time for server list + data move read
transaction and keyServer scan. Warn when getRange(dataMoveKeys) takes
over 5 seconds.
DataDistribution: Add NumShards and NumServers to DDInitGotInitialDD.
Add DDInitResumedDataMoves summary with ValidMoves, CancelledMoves,
EmptyMoves counts and elapsed time. Log DD exit reason as DDExiting
at SevWarn. Add DDInit-prefixed trace events throughout startup sequence.
DDTeamCollection: Add Reason and Address details to UndesiredStorageServer
trace events to distinguish version lag, same-address, wrong-class, and
exclusion causes.
DDShardTracker: Log TrackInitialShardsComplete with shard count and
TrackInitialShardsMetricsComplete with elapsed time.
NativeAPI: Add retry counting and logging for getStorageMetrics timeouts.
* Remove double trace
* Fix DD trace event issues: rate-limit warns, deduplicate events, fix DDExiting
- Wire up unused lastLogTime in waitStorageMetrics to rate-limit SevWarn
events to once per 10s after the 60s threshold. Previously every retry
(up to 100/s with 10ms WRONG_SHARD_SERVER_DELAY) emitted SevWarn.
- Remove redundant DataDistributorRunning and DDInitRunning events inside
the try block. Rename the original DataDistributorRunning to DDInitRunning
so a single event serves both purposes and fits the DDInit* query pattern.
- Move DDExiting to the three actual throw sites so it only fires on
terminal exits, not on retries of movekeys_conflict/dd_config_changed
when DD is disabled.
* Trim over-apologetic DDInitDone comment
* Add restore validation feature: restores to special keyspace allowing validating backup/restore in single cluster (space willing) (#12573)
* Add restore validation feature with simplified backup gap fix
Implements restore validation using audit_storage to verify backup/restore
correctness. Includes a minimal fix for the backup gap bug.
Key components:
- ValidateRestore audit type: compares source keys against restored keys
at \xff\x02/rlog/ prefix in storage server
- DD audit fixes: propagate validation errors, handle DD failover correctly
- RestoreValidation and BackupAndRestoreValidation workloads for testing
- Simplified backup gap fix: prevent snapshot from finishing in the same
iteration it dispatches the last tasks (single flag + one check)
* No faultInjection in 7.4
* Make it so no encryption when the restore validation test runs
* Add error_code_audit_storage_task_outdated to bypass list rather than do special-case handling
---------
Co-authored-by: michael stack <stack@duboce.com>
* Add multipart retry delay configuration. (#12076)
Add documentation to .h file.
Co-authored-by: stack <stack@duboce.com>
* Add a quickstart section on the bulkdump/bulkload feature to the user doc. (#12090)
* Fix assertion failure in fdbcli (#12095)
* Add Option to Turn Off Failure Injection to BulkLoad Process (#12096)
* add option to turn off failure injection to bulkload process in simulation
* nit
* DD BulkLoadTask Load Balance (#12103)
* bulkload load balance
* improve bulkload task submission efficiency
* address comments
* address comments
* cleanup
* Take a --proxy argument as backup_agent does. (#12110)
* Add "--proxy" to fdbserver (#12109)
* add --proxy to fdbserver
* fmt
* address comments
* use g_network to store proxy for s3client_ci
* Add 'ls' to s3client for the case where 'aws s3' is not installable (#12118)
* Add 'ls' to s3client for the case where 'aws s3' is not installable
-- i.e. production -- and we want a tool to test s3 connectivity.
Use it like this:
~/build_output/bin/s3client \
--tls-ca-file /etc/ssl/cert.pem \
--blob-credentials /path/to/credentials.json \
ls "blobstore://backup-us-west-2.s3.amazonaws.com/x?bucket=backup-us-west-2®ion=us-west-2"
* fdbclient/S3BlobStore.actor.cpp
* fdbclient/include/fdbclient/S3BlobStore.h
Make constructResourcePath method on endpoint so accessible making ls
URLs. Handling for encoded ampersands too. Allow 404 when requesting
resource that doesn't exist.
* fdbclient/S3Client.actor.cpp
* fdbclient/S3Client_cli.actor.cpp
Add in ls implementation.
* fdbclient/tests/s3client_test.sh
Add test for new ls facility.
* See if this fixes compile issue
---------
Co-authored-by: michael stack <stack@duboce.com>
* Ingest sst files rather than their keyvalue content (2nd attempt) (#12108)
* Ingest sst files rather than their keyvalue content.
* fdbclient/ServerKnobs.cpp
* fdbclient/include/fdbclient/ServerKnobs.h
Add BULK_LOAD_USE_SST_INGEST knob.
* fdbclient/include/fdbclient/IKeyValueStore.actor.h
Add ingestSSTFiles and supportSStIngestion.
* fdbserver/KeyValueStoreRocksDB.actor.cpp
* fdbserver/KeyValueStoreShardedRocksDB.actor.cpp
Implement ingestSSTFiles and supportSStIngestion
* fdbserver/storageserver.actor.cpp
If BULK_LOAD_USE_SST_INGEST and BulkLoadType::SST, ingest sst file
rather than read keyvalues.
* fdbclient/tests/fdb_cluster_fixture.sh
Use rocksdb instead of sqllite in tests.
* Revert "Ingest sst files rather than their keyvalue content."
This reverts commit 6d50bf15d3566de17fa6febf2afba0f220846ca6.
* ctests pass
* Restore ingest minus metric
* Add back ingestDuration metric
* Formatting
---------
Co-authored-by: michael stack <stack@duboce.com>
* Allow resource to be empty (#12129)
* RocksDB Do Compact After BulkLoad (#12130)
* rocksdb bulkload compact after load
* address comments
* address comments
* job overwrite task root path (#12138)
* BulkLoad Load Balance Based on DataMove Busy Map (#12137)
* bulkload load balance in ddqueue
* cleanup
* address comments
* nit
* bug fix
* nit
* address comments
* avoid scheduler busy loop
* address comments
* address comments
* avoid busy loop (#12157)
* First set of fixes... sizes and implement deleteResource (#12156)
Co-authored-by: michael stack <stack@duboce.com>
---------
Co-authored-by: Michael Stack <saintstack@users.noreply.github.com>
Co-authored-by: stack <stack@duboce.com>
* fdbcli/BulkDumpCommand.actor.cpp
* fdbcli/BulkLoadCommand.actor.cpp
Print out the bulkdump description rather than usage so user
has a chance of figuring out what it is they entered incorrectly.
Make bulkdump and bulkload align by using 'cancel' instead of
'clear' in both and ordering the sub-commands the same for
bulkload and bulkdump. Add more help to the description.
Bulkload was missing mention of the jobid needed
specifying a bulkload.
* documentation/sphinx/source/bulkdump.rst
s/clearBulkDumpJob/cancelBulkDumpJob/
Co-authored-by: stack <stack@duboce.com>
* * fdbclient/S3Client.actor.cpp
Change field names so capitialized (convention)
Add duration as field to traces.
* fdbserver/BulkLoadUtil.actor.cpp
When the job-manifest is big, processing blocks
so much getBulkLoadJobFileManifestEntryFromJobManifestFile
fails.
* Make bulkload file reads and writes async and memory parsimonious.
In tests at scale, processing a large job-manifest.txt was blocking
and causing the bulk job to fail. This is part 1 of two patches.
The second is to address data copy added in the below when we
made methods ACTORs (ACTOR doesn't allow passing by reference).
* fdbserver/BulkDumpUtil.actor.cpp
Removed writeStringToFile and buldDumpFileCopy in favor of new methods
in BulkLoadUtil. Made hosting functions ACTORs so could wait on
async calls.
* fdbserver/BulkLoadUtil.actor.cpp
Added async read and write functions.
* fdbserver/DataDistribution.actor.cpp
Making uploadBulkDumpJobManifestFile async made it so big bulkloads
work.
* fix memory corruption in writeBulkFileBytes and fix read options in getBulkLoadJobFileManifestEntryFromJobManifestFile
* If read or write < 1MB, do it in a single read else do multiple read/writes
* * packaging/docker/fdb-aws-s3-credentials-fetcher/fdb-aws-s3-credentials-fetcher.go
Just be blunt and write out the credentials. Trying to figure when the
blob credentials have expired is error prone.
Co-authored-by: michael stack <stack@duboce.com>
Co-authored-by: Zhe Wang <zhe.wang@wustl.edu>
* improve bulkload code
* address CI
* disable audit storage replica check and distributed consistency check in bulkload and bulkdump simulation test
* fix ci
* disable waitForQuiescence in bulkload and bulkdump tests
* bulkload support general engine and fix bugs
* add comments
* improve test coverage and fix bug
* nits and address comments
* nit
* nits
* fix data inconsistency bug due to bulkload metadata
* fix ss bulkload task metadata bugs
* nit and fix CI issue
* fix bugs of restore ss bulkload metadata
* use ssBulkLoadMetadata for fetchKey and general kv engine
* cleanup bulkload file for fetchkey
* fix CI issue
* fix simulation stuck due to repeated re-recruitment of unfit dd
* randomly do available space check when finding the dest team for bulkload in simulation
* address conflict
* code clean up
* update BulkDumping.toml same to BulkLoading.toml
* consolidate ss fetchkey and fetchshard failed to read bulkload task metadata
* fix DD bulkload job busy loop bug which causes segfault and test terminate unexpectedly in joshua test
* nit
* fix ss busy loop for bulkload in fetchkey
* use sqlite for bulkload ctest
* fix bulkload ctest stuck issue due to merge and change storage engine to ssd
* fix comments for CC recruit DD
* address comments
* address comments
* add comments
* fix ci format issue
* address comments
* add comments
It was disabled until we made it so the SS could
talk to s3, included in this PR.
Also finished the bulkload test. It only had the
bulkdump portion. bulkload support was recentlty
added so finish off the test here by adding bulkload
of the bulkdump and then verifying all data present.
Added passing knobs to the fdb cluster so available to the
fdbserver when it goes to talk to s3. Also added passing
SS count to start in fdb cluster.
* fdbclient/tests/fdb_cluster_fixture.sh
Add ability to pass multiple knobs to fdb cluster
and to specify more than just one SS.
* fdbserver/fdbserver.actor.cpp
Add --blob-server option and processing of FDB_BLOB_CREDENTIALS
if present (hijacked the unused, unadvertised --
blob-credentials-file).
* tests/loopback_cluster/run_custom_cluster.sh
Allow passing more than just one knob.
* fdbclient/BulkLoading.cpp
* fdbclient/include/fdbclient/BulkLoading.h
Added getPath
* fdbclient/S3BlobStore.actor.cpp
Fix bug where we were doubling up the first '/' on a path if
it had a root '/' already (s3 treats /a/b as distinct from
/a//b).
* fdbclient/S3Client.actor.cpp
Fix up of traceevent Types.
* fdbclient/tests/bulkload_test.sh
Enable being able to use s3 if available.
Pick up jobid when bulkdumping. Feed it to new bulkload
method. Add verification all data present post-bulkload.
* fdbserver/BulkLoadUtil.actor.cpp
Add support for blobstore.
* tests/loopback_cluster/run_custom_cluster.sh
Bug fix -- we were only able to pass in one knob. Allow
passing multiple.
* add bulkload job framework and fix bugs
* add BulkLoadChecksum, fix CI issue
* nits
* nits
* address comments
* mitigate perpetual wiggle to make sure DD can select a valid team to inject data
* fix submitBulkDumpJob and submitBulkLoadJob
* change remoteRoot to jobRoot
* add comments
* Improve BulkLoad/Dump implementation
* make bulkload test data folder inside simfdb folder
* simplify code
* use manifest in bulkdump metadata
* use manifest in bulkload
* apply bulkload fileset to bulkload and fix bugs of bytesampling value generation
* remove BulkDumpFileFullPathSet
* address comments
* address comments
* address comments
* init
* Add bulkdump to blobstore:// (s3)
* cmake/CompileBoost.cmake
Add boost url. Needed parsing blobstore:// urls.
* documentation/sphinx/source/bulkdump.rst
Minor edit to allow addition of blobstore target.
* fdbcli/BulkDumpCommand.actor.cpp
* fdbclient/BulkDumping.cpp
s/blobstore/s3/ -- more generic and aligns with
how backup/restore refers to "s3" thingies.
* fdbclient/include/fdbclient/S3Client.actor.h
* fdbclient/S3Client.actor.cpp
Add batch upload handler.
* fdbclient/tests/seaweedfs_fixture.sh
Add run seaweed method. Also look for
weed and if installed use it else download.
* fdbserver/BulkDumpUtil.actor.cpp
appendToPath does the right thing when passed an URL
Add bulkDumpTransportBlobstore_impl.
Add upload to blobstore.
* tests/loopback_cluster/run_custom_cluster.sh
Complain if unrecognized arguments.
* Add ctest for bulkload with simple bulkdump test for now.
* Add new test to ctest list
* fix bugs
* nit
* nits
* nits
---------
Co-authored-by: stack <stack@duboce.com>
michael stack