upgpkg: tikv 5.0.1-1

3 files changed, 698 insertions, 253 deletions

diff --git a/.SRCINFO b/.SRCINFO
index 715727c9f95a..327a49fe8334 100644
--- a/.SRCINFO
+++ b/.SRCINFO
@@ -1,6 +1,6 @@
 pkgbase = tikv
 	pkgdesc = Distributed transactional key-value database, originally created to complement TiDB
-	pkgver = 5.0.0
+	pkgver = 5.0.1
 	pkgrel = 1
 	url = https://github.com/tikv/tikv
 	arch = x86_64
@@ -15,16 +15,16 @@ pkgbase = tikv
 	depends = gcc-libs
 	provides = tikv-server
 	backup = etc/tikv/tikv.toml
-	source = tikv-5.0.0.tar.gz::https://github.com/tikv/tikv/archive/v5.0.0.tar.gz
+	source = tikv-5.0.1.tar.gz::https://github.com/tikv/tikv/archive/v5.0.1.tar.gz
 	source = tikv.service
 	source = tikv-sysusers.conf
 	source = tikv-tmpfiles.conf
 	source = tikv.toml
-	sha256sums = 2b14214184372f0d3c13a6be6f8bca8b6b4c93b34a9a6f5b5ebd836b63e97013
+	sha256sums = 15c5a7dc86d35da3ab350107e7142005b951b78bdc08eb9cdecb93716739b635
 	sha256sums = 870b8eaf83bc0d22b05b0f3a7890660e483cf77bb1d84bc50ad04fb23068cd8c
 	sha256sums = 744b252e29099b0099dc41e30bc3badd33b3d661c7126af8044faa4fc2df8927
 	sha256sums = 935291bac6a216c6f880df9bfaec8900266413bb202ac483e79f291e1f28e9f1
-	sha256sums = 24aff9d7aa90a18ac3a0474f4e4575baa5051a7ee0bb9a4c3a3d4aff49af7ac7
+	sha256sums = be2f8c6830a48da6c356db943aa55ee2f3c9c30b2e9027e7b758cab875fc8520
 
 pkgname = tikv
 
diff --git a/PKGBUILD b/PKGBUILD
index c26ed138e240..501e8c1df4a4 100644
--- a/PKGBUILD
+++ b/PKGBUILD
@@ -1,7 +1,7 @@
 # Maintainer: Xuanwo <xuanwo@archlinuxcn.org>
 # Maintainer: Allen Zhong <zhongbenli@pingcap.com>
 pkgname=tikv
-pkgver=5.0.0
+pkgver=5.0.1
 pkgrel=1
 pkgdesc='Distributed transactional key-value database, originally created to complement TiDB'
 makedepends=('go' 'make' 'rustup' 'awk' 'cmake' 'gcc')
@@ -16,11 +16,11 @@ source=(tikv-${pkgver}.tar.gz::https://github.com/tikv/tikv/archive/v${pkgver}.t
         tikv-sysusers.conf
         tikv-tmpfiles.conf
         tikv.toml)
-sha256sums=('2b14214184372f0d3c13a6be6f8bca8b6b4c93b34a9a6f5b5ebd836b63e97013'
+sha256sums=('15c5a7dc86d35da3ab350107e7142005b951b78bdc08eb9cdecb93716739b635'
             '870b8eaf83bc0d22b05b0f3a7890660e483cf77bb1d84bc50ad04fb23068cd8c'
             '744b252e29099b0099dc41e30bc3badd33b3d661c7126af8044faa4fc2df8927'
             '935291bac6a216c6f880df9bfaec8900266413bb202ac483e79f291e1f28e9f1'
-            '24aff9d7aa90a18ac3a0474f4e4575baa5051a7ee0bb9a4c3a3d4aff49af7ac7')
+            'be2f8c6830a48da6c356db943aa55ee2f3c9c30b2e9027e7b758cab875fc8520')
 
 prepare() {
     cd tikv-${pkgver}
diff --git a/tikv.toml b/tikv.toml
index e7741228bc9c..4d698ef5d303 100644
--- a/tikv.toml
+++ b/tikv.toml
@@ -1,41 +1,98 @@
-# TiKV config template
-#  Human-readable big numbers:
-#   File size(based on byte): KB, MB, GB, TB, PB
-#    e.g.: 1_048_576 = "1MB"
-#   Time(based on ms): ms, s, m, h
-#    e.g.: 78_000 = "1.3m"
-
-# log level: trace, debug, info, warning, error, critical.
-# Note that `debug` and `trace` are only available in development builds.
+## TiKV config template
+##  Human-readable big numbers:
+##   File size(based on byte): KB, MB, GB, TB, PB
+##    e.g.: 1_048_576 = "1MB"
+##   Time(based on ms): ms, s, m, h
+##    e.g.: 78_000 = "1.3m"
+
+## Log levels: trace, debug, info, warning, error, critical.
+## Note that `debug` and `trace` are only available in development builds.
 # log-level = "info"
 
-# file to store log, write to stderr if it's empty.
+## File to store logs.
+## If it is not set, logs will be appended to stderr.
 # log-file = ""
 
-# timespan between rotating the log files.
-# Once this timespan passes the existing log file will have a timestamp appended to its name,
-# and a new file will be created.
+## Log format, one of json, text. Default to text.
+# log-format = "text"
+
+## File to store slow logs.
+## If "log-file" is set, but this is not set, the slow logs will be appeneded
+## to "log-file". If both "log-file" and "slow-log-file" are not set, all logs
+## will be appended to stderr.
+# slow-log-file = ""
+
+## The minimum operation cost to output relative logs.
+# slow-log-threshold = "1s"
+
+## Timespan between rotating the log files.
+## Once this timespan passes, log files will be rotated, i.e. existing log file will have a
+## timestamp appended to its name and a new file will be created.
 # log-rotation-timespan = "24h"
 
+## Size of log file that triggers the log rotation.
+## Once the size of log file exceeds the threshold value, the log file will be rotated
+## and place the old log file in a new file named by orginal file name subbfixed by a timestamp.
+# log-rotation-size = "300MB"
+
+## Enable io snoop which utilize eBPF to get accurate disk io of TiKV
+## It won't take effect when compiling without BCC_IOSNOOP=1. 
+# enable-io-snoop = true
+
+# Configurations for the single thread pool serving read requests.
+[readpool.unified]
+## The minimal working thread count of the thread pool.
+# min-thread-count = 1
+
+## The maximum working thread count of the thread pool.
+## The default value is max(4, LOGICAL_CPU_NUM * 0.8).
+# max-thread-count = 8
+
+## Size of the stack for each thread in the thread pool.
+# stack-size = "10MB"
+
+## Max running tasks of each worker, reject if exceeded.
+# max-tasks-per-worker = 2000
+
 [readpool.storage]
-# size of thread pool for high-priority operations
+## Whether to use the unified read pool to handle storage requests.
+# use-unified-pool = false
+
+## The following configurations only take effect when `use-unified-pool` is false.
+
+## Size of the thread pool for high-priority operations.
 # high-concurrency = 4
-# size of thread pool for normal-priority operations
+
+## Size of the thread pool for normal-priority operations.
 # normal-concurrency = 4
-# size of thread pool for low-priority operations
+
+## Size of the thread pool for low-priority operations.
 # low-concurrency = 4
-# max running high-priority operations of each worker, reject if exceed
+
+## Max running high-priority operations of each worker, reject if exceeded.
 # max-tasks-per-worker-high = 2000
-# max running normal-priority operations of each worker, reject if exceed
+
+## Max running normal-priority operations of each worker, reject if exceeded.
 # max-tasks-per-worker-normal = 2000
-# max running low-priority operations of each worker, reject if exceed
+
+## Max running low-priority operations of each worker, reject if exceeded.
 # max-tasks-per-worker-low = 2000
-# size of stack size for each thread pool
+
+## Size of the stack for each thread in the thread pool.
 # stack-size = "10MB"
 
 [readpool.coprocessor]
-# Notice: if CPU_NUM > 8, default thread pool size for coprocessors
-# will be set to CPU_NUM * 0.8.
+## Whether to use the unified read pool to handle coprocessor requests.
+# use-unified-pool = true
+
+## The following configurations only take effect when `use-unified-pool` is false.
+
+## Most read requests from TiDB are sent to the coprocessor of TiKV. high/normal/low-concurrency is
+## used to set the number of threads of the coprocessor.
+## If there are many read requests, you can increase these config values (but keep it within the
+## number of system CPU cores). For example, for a 32-core machine deployed with TiKV, you can even
+## set these config to 30 in heavy read scenarios.
+## If CPU_NUM > 8, the default thread pool size for coprocessors is set to CPU_NUM * 0.8.
 
 # high-concurrency = 8
 # normal-concurrency = 8
@@ -43,20 +100,19 @@
 # max-tasks-per-worker-high = 2000
 # max-tasks-per-worker-normal = 2000
 # max-tasks-per-worker-low = 2000
-# stack-size = "10MB"
 
 [server]
-# set listening address.
+## Listening address.
 # addr = "127.0.0.1:20160"
-# set advertise listening address for client communication, if not set, use addr instead.
+
+## Advertise listening address for client communication.
+## If not set, `addr` will be used.
 # advertise-addr = ""
-# notify capacity, 40960 is suitable for about 7000 regions.
-# notify-capacity = 40960
-# maximum number of messages can be processed in one tick.
-# messages-per-tick = 4096
 
 ## Status address.
-## This is used for reporting the status of TiKV directly through the HTTP address.
+## This is used for reporting the status of TiKV directly through 
+## the HTTP address. Notice that there is a risk of leaking status
+## information if this port is exposed to the public.
 ## Empty string means disabling it.
 # status-addr = "127.0.0.1:20180"
 
@@ -65,376 +121,628 @@
 
 ## Compression type for gRPC channel: none, deflate or gzip.
 # grpc-compression-type = "none"
-# size of thread pool for grpc server.
+
+## Size of the thread pool for the gRPC server.
 # grpc-concurrency = 4
-# The number of max concurrent streams/requests on a client connection.
+
+## The number of max concurrent streams/requests on a client connection.
 # grpc-concurrent-stream = 1024
-# The number of connections with each tikv server to send raft messages.
-# grpc-raft-conn-num = 10
-# Amount to read ahead on individual grpc streams.
+
+## Limit the memory size can be used by gRPC. Default is unlimited.
+## gRPC usually works well to reclaim memory by itself. Limit the memory in case OOM
+## is observed. Note that limit the usage can lead to potential stall.
+# grpc-memory-pool-quota = "32G"
+
+## The number of connections with each TiKV server to send Raft messages.
+# grpc-raft-conn-num = 1
+
+## Amount to read ahead on individual gRPC streams.
 # grpc-stream-initial-window-size = "2MB"
-# Time to wait before sending out a ping to check if server is still alive.
-# This is only for communications between tikv instances.
+
+## Time to wait before sending out a ping to check if server is still alive.
+## This is only for communications between TiKV instances.
 # grpc-keepalive-time = "10s"
-# Time to wait before closing the connection without receiving keepalive ping
-# ack.
+
+## Time to wait before closing the connection without receiving KeepAlive ping Ack.
 # grpc-keepalive-timeout = "3s"
 
-# How many snapshots can be sent concurrently.
+## How many snapshots can be sent concurrently.
 # concurrent-send-snap-limit = 32
-# How many snapshots can be recv concurrently.
+
+## How many snapshots can be received concurrently.
 # concurrent-recv-snap-limit = 32
 
-# max recursion level allowed when decoding dag expression
+## Max allowed recursion level when decoding Coprocessor DAG expression.
 # end-point-recursion-limit = 1000
 
-# max time to handle coprocessor request before timeout
+## Max time to handle Coprocessor requests before timeout.
 # end-point-request-max-handle-duration = "60s"
 
-# the max bytes that snapshot can be written to disk in one second,
-# should be set based on your disk performance
+## Max bytes that snapshot can be written to disk in one second.
+## It should be set based on your disk performance.
 # snap-max-write-bytes-per-sec = "100MB"
 
-# set attributes about this server, e.g. { zone = "us-west-1", disk = "ssd" }.
+## Whether to enable request batch.
+# enable-request-batch = true
+
+## Attributes about this server, e.g. `{ zone = "us-west-1", disk = "ssd" }`.
 # labels = {}
 
-[storage]
-# set the path to rocksdb directory.
-data-dir = "/var/lib/tikv/store"
+## The working thread count of the background pool, which include the endpoint of and br, split-check,
+## region thread and other thread of delay-insensitive tasks.
+## The default value is 2 if the number of CPU cores is less than 16, otherwise 3.
+# background-thread-count = 2
+
+## If handle time is larger than the threshold, it will print slow log in endpoint.
+## The default value is 1s.
+# end-point-slow-log-threshold = "1s"
 
-# notify capacity of scheduler's channel
-# scheduler-notify-capacity = 10240
+[storage]
+## The path to RocksDB directory.
+# data-dir = "/tmp/tikv/store"
 
-# the number of slots in scheduler latches, concurrency control for write.
+## The number of slots in Scheduler latches, which controls write concurrency.
+## In most cases you can use the default value. When importing data, you can set it to a larger
+## value.
 # scheduler-concurrency = 2048000
 
-# scheduler's worker pool size, should increase it in heavy write cases,
-# also should less than total cpu cores.
+## Scheduler's worker pool size, i.e. the number of write threads.
+## It should be less than total CPU cores. When there are frequent write operations, set it to a
+## higher value. More specifically, you can run `top -H -p tikv-pid` to check whether the threads
+## named `sched-worker-pool` are busy.
 # scheduler-worker-pool-size = 4
 
-# When the pending write bytes exceeds this threshold,
-# the "scheduler too busy" error is displayed.
+## When the pending write bytes exceeds this threshold, the "scheduler too busy" error is displayed.
 # scheduler-pending-write-threshold = "100MB"
 
+## For async commit transactions, it's possible to response to the client before applying prewrite
+## requests. Enabling this can ease reduce latency when apply duration is significant, or reduce
+## latency jittering when apply duration is not stable.
+# enable-async-apply-prewrite = false
+
+## Reserve some space to ensure recovering the store from `no space left` must succeed.
+## `max(reserve-space, capacity * 5%)` will be reserved exactly.
+##
+## Set it to 0 will cause no space is reserved at all. It's generally used for tests.
+# reserve-space = "5GB"
+
+[storage.block-cache]
+## Whether to create a shared block cache for all RocksDB column families.
+##
+## Block cache is used by RocksDB to cache uncompressed blocks. Big block cache can speed up read.
+## It is recommended to turn on shared block cache. Since only the total cache size need to be
+## set, it is easier to config. In most cases it should be able to auto-balance cache usage
+## between column families with standard LRU algorithm.
+##
+## The rest of config in the storage.block-cache session is effective only when shared block cache
+## is on.
+# shared = true
+
+## Size of the shared block cache. Normally it should be tuned to 30%-50% of system's total memory.
+## When the config is not set, it is decided by the sum of the following fields or their default
+## value:
+##   * rocksdb.defaultcf.block-cache-size or 25% of system's total memory
+##   * rocksdb.writecf.block-cache-size   or 15% of system's total memory
+##   * rocksdb.lockcf.block-cache-size    or  2% of system's total memory
+##   * raftdb.defaultcf.block-cache-size  or  2% of system's total memory
+##
+## To deploy multiple TiKV nodes on a single physical machine, configure this parameter explicitly.
+## Otherwise, the OOM problem might occur in TiKV.
+# capacity = "1GB"
+
 [pd]
-# pd endpoints
-endpoints = ["127.0.0.1:2379"]
+## PD endpoints.
+# endpoints = []
 
-[raftstore]
-# true (default value) for high reliability, this can prevent data loss when power failure.
-# sync-log = true
+## The interval at which to retry a PD connection initialization.
+## Default is 300ms.
+# retry-interval = "300ms"
+
+## If the client observes an error, it can can skip reporting it except every `n` times.
+## Set to 1 to disable this feature.
+## Default is 10.
+# retry-log-every = 10
+
+## The maximum number of times to retry a PD connection initialization.
+## Set to 0 to disable retry.
+## Default is -1, meaning isize::MAX times.
+# retry-max-count = -1
 
-# minimizes disruption when a partitioned node rejoins the cluster by using a two phase election.
+[raftstore]
+## Whether to enable Raft prevote.
+## Prevote minimizes disruption when a partitioned node rejoins the cluster by using a two phase
+## election.
 # prevote = true
 
-# set the path to raftdb directory, default value is data-dir/raft
+## The path to RaftDB directory.
+## If not set, it will be `{data-dir}/raft`.
+## If there are multiple disks on the machine, storing the data of Raft RocksDB on differen disks
+## can improve TiKV performance.
 # raftdb-path = ""
 
-# set store capacity, if no set, use disk capacity.
+## Store capacity, i.e. max data size allowed.
+## If it is not set, disk capacity is used.
 # capacity = 0
 
-# notify capacity, 40960 is suitable for about 7000 regions.
+## Internal notify capacity.
+## 40960 is suitable for about 7000 Regions. It is recommended to use the default value.
 # notify-capacity = 40960
 
-# maximum number of messages can be processed in one tick.
+## Maximum number of internal messages to process in a tick.
 # messages-per-tick = 4096
 
-# Region heartbeat tick interval for reporting to pd.
+## Region heartbeat tick interval for reporting to PD.
 # pd-heartbeat-tick-interval = "60s"
-# Store heartbeat tick interval for reporting to pd.
+
+## Store heartbeat tick interval for reporting to PD.
 # pd-store-heartbeat-tick-interval = "10s"
 
-# When region size changes exceeds region-split-check-diff, we should check
-# whether the region should be split or not.
+## The threshold of triggering Region split check.
+## When Region size change exceeds this config, TiKV will check whether the Region should be split
+## or not. To reduce the cost of scanning data in the checking process, you can set the value to
+## 32MB during checking and set it back to the default value in normal operations.
 # region-split-check-diff = "6MB"
 
-# Interval to check region whether need to be split or not.
+## The interval of triggering Region split check.
 # split-region-check-tick-interval = "10s"
 
-# When raft entry exceed the max size, reject to propose the entry.
+## When the number of Raft entries exceeds the max size, TiKV rejects to propose the entry.
 # raft-entry-max-size = "8MB"
 
-# Interval to gc unnecessary raft log.
+## Interval to GC unnecessary Raft log.
 # raft-log-gc-tick-interval = "10s"
-# A threshold to gc stale raft log, must >= 1.
+
+## Threshold to GC stale Raft log, must be >= 1.
 # raft-log-gc-threshold = 50
-# When entry count exceed this value, gc will be forced trigger.
+
+## When the entry count exceeds this value, GC will be forced to trigger.
 # raft-log-gc-count-limit = 72000
-# When the approximate size of raft log entries exceed this value, gc will be forced trigger.
-# It's recommanded to set it to 3/4 of region-split-size.
+
+## When the approximate size of Raft log entries exceeds this value, GC will be forced trigger.
+## It's recommanded to set it to 3/4 of `region-split-size`.
 # raft-log-gc-size-limit = "72MB"
 
-# When a peer hasn't been active for max-peer-down-duration,
-# we will consider this peer to be down and report it to pd.
+## Raft engine is a replaceable component. For some implementations, it's necessary to purge
+## old log files to recycle disk space ASAP.
+# raft-engine-purge-interval = "10s"
+
+## How long the peer will be considered down and reported to PD when it hasn't been active for this
+## time.
 # max-peer-down-duration = "5m"
 
-# Interval to check whether start manual compaction for a region,
+## Interval to check whether to start manual compaction for a Region.
 # region-compact-check-interval = "5m"
 
-# Number of regions for each time to check.
+## Number of Regions for each time to check.
 # region-compact-check-step = 100
 
-# The minimum number of delete tombstones to trigger manual compaction.
+## The minimum number of delete tombstones to trigger manual compaction.
 # region-compact-min-tombstones = 10000
 
-# The minimum percentage of delete tombstones to trigger manual compaction.
-# Should between 1 and 100. Manual compaction only triggered when the number
-# of delete tombstones exceeds region-compact-min-tombstones and the percentage
-# of delete tombstones exceeds region-compact-tombstones-percent.
+## The minimum percentage of delete tombstones to trigger manual compaction.
+## It should be set between 1 and 100. Manual compaction is only triggered when the number of
+## delete tombstones exceeds `region-compact-min-tombstones` and the percentage of delete tombstones
+## exceeds `region-compact-tombstones-percent`.
 # region-compact-tombstones-percent = 30
 
-# Interval to check whether should start a manual compaction for lock column family,
-# if written bytes reach lock-cf-compact-threshold for lock column family, will fire
-# a manual compaction for lock column family.
+## Interval to check whether to start a manual compaction for Lock Column Family.
+## If written bytes reach `lock-cf-compact-bytes-threshold` for Lock Column Family, TiKV will
+## trigger a manual compaction for Lock Column Family.
 # lock-cf-compact-interval = "10m"
-
 # lock-cf-compact-bytes-threshold = "256MB"
 
-# Interval (s) to check region whether the data are consistent.
+## Interval (s) to check Region whether the data are consistent.
 # consistency-check-interval = 0
 
-# Use delete range to drop a large number of continuous keys.
-# use-delete-range = false
+## Interval to clean up import SST files.
+# cleanup-import-sst-interval = "10m"
 
-# delay time before deleting a stale peer
-# clean-stale-peer-delay = "10m"
+## Use how many threads to handle log apply
+# apply-pool-size = 2
 
-# Interval to cleanup import sst files.
-# cleanup-import-sst-interval = "10m"
+## Use how many threads to handle raft messages
+# store-pool-size = 2
 
 [coprocessor]
-# When it is true, it will try to split a region with table prefix if
-# that region crosses tables. It is recommended to turn off this option
-# if there will be a large number of tables created.
-# split-region-on-table = true
-# When the region's size exceeds region-max-size, we will split the region
-# into two which the left region's size will be region-split-size or a little
-# bit smaller.
+## When it is set to `true`, TiKV will try to split a Region with table prefix if that Region
+## crosses tables.
+## It is recommended to turn off this option if there will be a large number of tables created.
+# split-region-on-table = false
+
+## One split check produces several split keys in batch. This config limits the number of produced
+## split keys in one batch.
+# batch-split-limit = 10
+
+## When Region [a,e) size exceeds `region_max_size`, it will be split into several Regions [a,b),
+## [b,c), [c,d), [d,e) and the size of [a,b), [b,c), [c,d) will be `region_split_size` (or a
+## little larger).
 # region-max-size = "144MB"
 # region-split-size = "96MB"
-# When the region's keys exceeds region-max-keys, we will split the region
-# into two which the left region's keys will be region-split-keys or a little
-# bit smaller.
+
+## When the number of keys in Region [a,e) exceeds the `region_max_keys`, it will be split into
+## several Regions [a,b), [b,c), [c,d), [d,e) and the number of keys in [a,b), [b,c), [c,d) will be
+## `region_split_keys`.
 # region-max-keys = 1440000
 # region-split-keys = 960000
 
+## Set to "mvcc" to do consistency check for MVCC data, or "raw" for raw data.
+# consistency-check-method = "mvcc"
+
 [rocksdb]
-# Maximum number of concurrent background jobs (compactions and flushes)
+## Maximum number of threads of RocksDB background jobs.
+## The background tasks include compaction and flush. For detailed information why RocksDB needs to
+## do compaction, see RocksDB-related materials.
+## When write traffic (like the importing data size) is big, it is recommended to enable more
+## threads. But set the number of the enabled threads smaller than that of CPU cores. For example,
+## when importing data, for a machine with a 32-core CPU, set the value to 28.
+## The default value is set to 8 or CPU_NUM - 1, whichever is smaller.
 # max-background-jobs = 8
 
-# This value represents the maximum number of threads that will concurrently perform a
-# compaction job by breaking it into multiple, smaller ones that are run simultaneously.
-# Default: 1 (i.e. no subcompactions)
-# max-sub-compactions = 1
-
-# Number of open files that can be used by the DB.  You may need to
-# increase this if your database has a large working set. Value -1 means
-# files opened are always kept open. You can estimate number of files based
-# on target_file_size_base and target_file_size_multiplier for level-based
-# compaction.
-# If max-open-files = -1, RocksDB will prefetch index and filter blocks into
-# block cache at startup, so if your database has a large working set, it will
-# take several minutes to open the db.
+## Maximum number of threads of RocksDB background memtable flush jobs.
+## The default value is set to 2 or max_background_jobs / 4, whichever is bigger.
+# max-background-flushes = 2
+
+## Represents the maximum number of threads that will concurrently perform a sub-compaction job by
+## breaking it into multiple, smaller ones running simultaneously.
+## The default value is set to 3 or the largest number to allow for two compactions, whichever is
+## smaller.
+# max-sub-compactions = 3
+
+## Number of open files that can be used by the DB.
+## Value -1 means files opened are always kept open and RocksDB will prefetch index and filter
+## blocks into block cache at startup. So if your database has a large working set, it will take
+## several minutes to open the DB. You may need to increase this if your database has a large
+## working set. You can estimate the number of files based on `target-file-size-base` and
+## `target_file_size_multiplier` for level-based compaction.
 # max-open-files = 40960
 
-# Max size of rocksdb's MANIFEST file.
-# For detailed explanation please refer to https://github.com/facebook/rocksdb/wiki/MANIFEST
+## Max size of RocksDB's MANIFEST file.
+## For detailed explanation, please refer to https://github.com/facebook/rocksdb/wiki/MANIFEST
 # max-manifest-file-size = "128MB"
 
-# If true, the database will be created if it is missing.
+## If the value is `true`, the database will be created if it is missing.
 # create-if-missing = true
 
-# rocksdb wal recovery mode
-# 0 : TolerateCorruptedTailRecords, tolerate incomplete record in trailing data on all logs;
-# 1 : AbsoluteConsistency, We don't expect to find any corruption in the WAL;
-# 2 : PointInTimeRecovery, Recover to point-in-time consistency;
-# 3 : SkipAnyCorruptedRecords, Recovery after a disaster;
+## RocksDB Write-Ahead Logs (WAL) recovery mode.
+## 0 : TolerateCorruptedTailRecords, tolerate incomplete record in trailing data on all logs;
+## 1 : AbsoluteConsistency, We don't expect to find any corruption in the WAL;
+## 2 : PointInTimeRecovery, Recover to point-in-time consistency;
+## 3 : SkipAnyCorruptedRecords, Recovery after a disaster;
 # wal-recovery-mode = 2
 
-# rocksdb write-ahead logs dir path
-# This specifies the absolute dir path for write-ahead logs (WAL).
-# If it is empty, the log files will be in the same dir as data.
-# When you set the path to rocksdb directory in memory like in /dev/shm, you may want to set
-# wal-dir to a directory on a persistent storage.
-# See https://github.com/facebook/rocksdb/wiki/How-to-persist-in-memory-RocksDB-database
+## RocksDB WAL directory.
+## This config specifies the absolute directory path for WAL.
+## If it is not set, the log files will be in the same directory as data. When you set the path to
+## RocksDB directory in memory like in `/dev/shm`, you may want to set`wal-dir` to a directory on a
+## persistent storage. See https://github.com/facebook/rocksdb/wiki/How-to-persist-in-memory-RocksDB-database .
+## If there are two disks on the machine, storing RocksDB data and WAL logs on different disks can
+## improve performance.
 # wal-dir = "/tmp/tikv/store"
 
-# The following two fields affect how archived write-ahead logs will be deleted.
-# 1. If both set to 0, logs will be deleted asap and will not get into the archive.
-# 2. If wal-ttl-seconds is 0 and wal-size-limit is not 0,
-#    WAL files will be checked every 10 min and if total size is greater
-#    then wal-size-limit, they will be deleted starting with the
-#    earliest until size_limit is met. All empty files will be deleted.
-# 3. If wal-ttl-seconds is not 0 and wal-size-limit is 0, then
-#    WAL files will be checked every wal-ttl-seconds / 2 and those that
-#    are older than wal-ttl-seconds will be deleted.
-# 4. If both are not 0, WAL files will be checked every 10 min and both
-#    checks will be performed with ttl being first.
-# When you set the path to rocksdb directory in memory like in /dev/shm, you may want to set
-# wal-ttl-seconds to a value greater than 0 (like 86400) and backup your db on a regular basis.
-# See https://github.com/facebook/rocksdb/wiki/How-to-persist-in-memory-RocksDB-database
+## The following two fields affect how archived WAL will be deleted.
+## 1. If both values are set to 0, logs will be deleted ASAP and will not get into the archive.
+## 2. If `wal-ttl-seconds` is 0 and `wal-size-limit` is not 0, WAL files will be checked every 10
+##    min and if total size is greater than `wal-size-limit`, they will be deleted starting with the
+##    earliest until `wal-size-limit` is met. All empty files will be deleted.
+## 3. If `wal-ttl-seconds` is not 0 and `wal-size-limit` is 0, then WAL files will be checked every
+##    `wal-ttl-seconds / 2` and those that are older than `wal-ttl-seconds` will be deleted.
+## 4. If both are not 0, WAL files will be checked every 10 min and both checks will be performed
+##    with ttl being first.
+## When you set the path to RocksDB directory in memory like in `/dev/shm`, you may want to set
+## `wal-ttl-seconds` to a value greater than 0 (like 86400) and backup your DB on a regular basis.
+## See https://github.com/facebook/rocksdb/wiki/How-to-persist-in-memory-RocksDB-database .
 # wal-ttl-seconds = 0
 # wal-size-limit = 0
 
-# rocksdb max total wal size
+## Max RocksDB WAL size in total
 # max-total-wal-size = "4GB"
 
-# Rocksdb Statistics provides cumulative stats over time.
-# Turn statistics on will introduce about 5%-10% overhead for RocksDB,
-# but it is worthy to know the internal status of RocksDB.
+## RocksDB Statistics provides cumulative stats over time.
+## Turning statistics on will introduce about 5%-10% overhead for RocksDB, but it can help you to
+## know the internal status of RocksDB.
 # enable-statistics = true
 
-# Dump statistics periodically in information logs.
-# Same as rocksdb's default value (10 min).
+## Dump statistics periodically in information logs.
+## Same as RocksDB's default value (10 min).
 # stats-dump-period = "10m"
 
-# Due to Rocksdb FAQ: https://github.com/facebook/rocksdb/wiki/RocksDB-FAQ,
-# If you want to use rocksdb on multi disks or spinning disks, you should set value at
-# least 2MB;
+## Refer to: https://github.com/facebook/rocksdb/wiki/RocksDB-FAQ
+## If you want to use RocksDB on multi disks or spinning disks, you should set value at least 2MB.
 # compaction-readahead-size = 0
 
-# This is the maximum buffer size that is used by WritableFileWrite
+## Max buffer size that is used by WritableFileWrite.
 # writable-file-max-buffer-size = "1MB"
 
-# Use O_DIRECT for both reads and writes in background flush and compactions
+## Use O_DIRECT for both reads and writes in background flush and compactions.
 # use-direct-io-for-flush-and-compaction = false
 
-# Limit the disk IO of compaction and flush. Compaction and flush can cause
-# terrible spikes if they exceed a certain threshold. Consider setting this to
-# 50% ~ 80% of the disk throughput for a more stable result. However, in heavy
-# write workload, limiting compaction and flush speed can cause write stalls too.
-# rate-bytes-per-sec = 0
-
-# Enable or disable the pipelined write
+## Limit the disk IO of compaction and flush.
+## Compaction and flush can cause terrible spikes if they exceed a certain threshold. Consider
+## setting this to 50% ~ 80% of the disk throughput for a more stable result. However, in heavy
+## write workload, limiting compaction and flush speed can cause write stalls too.
+## 1. rate-bytes-per-sec is the only parameter you want to set most of the time. It controls the
+##    total write rate of compaction and flush in bytes per second. Currently, RocksDB does not
+##    enforce rate limit for anything other than flush and compaction, e.g. write to WAL.
+## 2. rate-limiter-refill-period controls how often IO tokens are refilled. Smaller value will flatten
+##    IO bursts while introducing more CPU overhead.
+## 3. rate-limiter-mode indicates which types of operations count against the limit.
+##    1 : ReadOnly
+##    2 : WriteOnly
+##    3 : AllIo
+## 4. rate-limiter-auto_tuned enables dynamic adjustment of rate limit within the range
+##    [10MB/s, rate_bytes_per_sec], according to the recent demand for background I/O.
+# rate-bytes-per-sec = "10GB"
+# rate-limiter-refill-period = "100ms"
+# rate-limiter-mode = 2
+# rate-limiter-auto-tuned = true
+
+## Enable or disable the pipelined write.
 # enable-pipelined-write = true
 
-# Allows OS to incrementally sync files to disk while they are being
-# written, asynchronously, in the background.
+## Allows OS to incrementally sync files to disk while they are being written, asynchronously,
+## in the background.
 # bytes-per-sync = "1MB"
 
-# Allows OS to incrementally sync WAL to disk while it is being written.
+## Allows OS to incrementally sync WAL to disk while it is being written.
 # wal-bytes-per-sync = "512KB"
 
-# Specify the maximal size of the Rocksdb info log file. If the log file
-# is larger than `max_log_file_size`, a new info log file will be created.
-# If max_log_file_size == 0, all logs will be written to one log file.
+## Specify the maximal size of the RocksDB info log file.
+## If the log file is larger than this config, a new info log file will be created.
+## If it is set to 0, all logs will be written to one log file.
 # info-log-max-size = "1GB"
 
-# Time for the Rocksdb info log file to roll (in seconds).
-# If specified with non-zero value, log file will be rolled
-# if it has been active longer than `log_file_time_to_roll`.
-# 0 means disabled.
-# info-log-roll-time = "0"
+## Time for the RocksDB info log file to roll (in seconds).
+## If the log file has been active longer than this config, it will be rolled.
+## If it is set to 0, rolling will be disabled.
+# info-log-roll-time = "0s"
 
-# Maximal Rocksdb info log files to be kept.
+## Maximal RocksDB info log files to be kept.
 # info-log-keep-log-file-num = 10
 
-# This specifies the Rocksdb info LOG dir.
-# If it is empty, the log files will be in the same dir as data.
-# If it is non empty, the log files will be in the specified dir,
-# and the db data dir's absolute path will be used as the log file
-# name's prefix.
+## Specifies the RocksDB info log directory.
+## If it is empty, the log files will be in the same directory as data.
+## If it is not empty, the log files will be in the specified directory, and the DB data directory's
+## absolute path will be used as the log file name's prefix.
 # info-log-dir = ""
 
-# Column Family default used to store actual data of the database.
+# RocksDB log levels
+# info-log-level = "info"
+
+## Options for `Titan`.
+[rocksdb.titan]
+## Enables or disables `Titan`. Note that Titan is still an experimental feature. Once
+## enabled, it can't fall back. Forced fallback may result in data loss.
+## default: false
+# enabled = false
+
+## Maximum number of threads of `Titan` background gc jobs.
+# default: 4
+# max-background-gc = 4
+
+## Options for "Default" Column Family, which stores actual user data.
 [rocksdb.defaultcf]
-# compression method (if any) is used to compress a block.
-#   no:     kNoCompression
-#   snappy: kSnappyCompression
-#   zlib:   kZlibCompression
-#   bzip2:  kBZip2Compression
-#   lz4:    kLZ4Compression
-#   lz4hc:  kLZ4HCCompression
-#   zstd:   kZSTD
-
-# per level compression
+## Compression method (if any) is used to compress a block.
+##   no:     kNoCompression
+##   snappy: kSnappyCompression
+##   zlib:   kZlibCompression
+##   bzip2:  kBZip2Compression
+##   lz4:    kLZ4Compression
+##   lz4hc:  kLZ4HCCompression
+##   zstd:   kZSTD
+## `lz4` is a compression algorithm with moderate speed and compression ratio. The compression
+## ratio of `zlib` is high. It is friendly to the storage space, but its compression speed is
+## slow. This compression occupies many CPU resources.
+
+## Per level compression.
+## This config should be chosen carefully according to CPU and I/O resources. For example, if you
+## use the compression mode of "no:no:lz4:lz4:lz4:zstd:zstd" and find much I/O pressure of the
+## system (run the `iostat` command to find %util lasts 100%, or run the `top` command to find many
+## iowaits) when writing (importing) a lot of data while the CPU resources are adequate, you can
+## compress level-0 and level-1 and exchange CPU resources for I/O resources. If you use the
+## compression mode of "no:no:lz4:lz4:lz4:zstd:zstd" and you find the I/O pressure of the system is
+## not big when writing a lot of data, but CPU resources are inadequate. Then run the `top` command
+## and choose the `-H` option. If you find a lot of bg threads (namely the compression thread of
+## RocksDB) are running, you can exchange I/O resources for CPU resources and change the compression
+## mode to "no:no:no:lz4:lz4:zstd:zstd". In a word, it aims at making full use of the existing
+## resources of the system and improving TiKV performance in terms of the current resources.
 # compression-per-level = ["no", "no", "lz4", "lz4", "lz4", "zstd", "zstd"]
 
-# Approximate size of user data packed per block.  Note that the
-# block size specified here corresponds to uncompressed data.
+## Set zstd compression for the bottommost level.
+## This config overrides compression-per-level. It use zstd for bottommost level to ensure
+## consistent compression ratio, regardless of overall data size. If explicitly setting compression type
+## for the bottommost level is not desired, "disable" should be used.
+# bottommost-level-compression = "zstd"
+
+## The data block size. RocksDB compresses data based on the unit of block.
+## Similar to page in other databases, block is the smallest unit cached in block-cache. Note that
+## the block size specified here corresponds to uncompressed data.
 # block-size = "64KB"
 
-# If you're doing point lookups you definitely want to turn bloom filters on, We use
-# bloom filters to avoid unnecessary disk reads. Default bits_per_key is 10, which
-# yields ~1% false positive rate. Larger bits_per_key values will reduce false positive
-# rate, but increase memory usage and space amplification.
+## If you're doing point lookups you definitely want to turn bloom filters on. We use bloom filters
+## to avoid unnecessary disk reads. Default bits_per_key is 10, which yields ~1% false positive
+## rate. Larger `bloom-filter-bits-per-key` values will reduce false positive rate, but increase
+## memory usage and space amplification.
 # bloom-filter-bits-per-key = 10
 
-# false means one sst file one bloom filter, true means evry block has a corresponding bloom filter
+## `false` means one SST file one bloom filter, `true` means every block has a corresponding bloom
+## filter.
 # block-based-bloom-filter = false
 
 # level0-file-num-compaction-trigger = 4
 
-# Soft limit on number of level-0 files. We start slowing down writes at this point.
+## Soft limit on number of level-0 files.
+## When the number of SST files of level-0 reaches the limit of `level0-slowdown-writes-trigger`,
+## RocksDB tries to slow down the write operation, because too many SST files of level-0 can cause
+## higher read pressure of RocksDB.
 # level0-slowdown-writes-trigger = 20
 
-# Maximum number of level-0 files.  We stop writes at this point.
+## Maximum number of level-0 files.
+## When the number of SST files of level-0 reaches the limit of `level0-stop-writes-trigger`,
+## RocksDB stalls the new write operation.
 # level0-stop-writes-trigger = 36
 
-# Amount of data to build up in memory (backed by an unsorted log
-# on disk) before converting to a sorted on-disk file.
+## Amount of data to build up in memory (backed by an unsorted log on disk) before converting to a
+## sorted on-disk file. It is the RocksDB MemTable size.
 # write-buffer-size = "128MB"
 
-# The maximum number of write buffers that are built up in memory.
+## The maximum number of the MemTables. The data written into RocksDB is first recorded in the WAL
+## log, and then inserted into MemTables. When the MemTable reaches the size limit of
+## `write-buffer-size`, it turns into read only and generates a new MemTable receiving new write
+## operations. The flush threads of RocksDB will flush the read only MemTable to the disks to become
+## an SST file of level0. `max-background-flushes` controls the maximum number of flush threads.
+## When the flush threads are busy, resulting in the number of the MemTables waiting to be flushed
+## to the disks reaching the limit of `max-write-buffer-number`, RocksDB stalls the new operation.
+## "Stall" is a flow control mechanism of RocksDB. When importing data, you can set the
+## `max-write-buffer-number` value higher, like 10.
 # max-write-buffer-number = 5
 
-# The minimum number of write buffers that will be merged together
-# before writing to storage.
+## The minimum number of write buffers that will be merged together before writing to storage.
 # min-write-buffer-number-to-merge = 1
 
-# Control maximum total data size for base level (level 1).
+## Control maximum total data size for base level (level 1).
+## When the level-1 data size reaches the limit value of `max-bytes-for-level-base`, the SST files
+## of level-1 and their overlap SST files of level-2 will be compacted. The golden rule: the first
+## reference principle of setting `max-bytes-for-level-base` is guaranteeing that the
+## `max-bytes-for-level-base` value is roughly equal to the data volume of level-0. Thus
+## unnecessary compaction is reduced. For example, if the compression mode is
+## "no:no:lz4:lz4:lz4:lz4:lz4", the `max-bytes-for-level-base` value can be `write-buffer-size * 4`,
+## because there is no compression of level-0 and level-1 and the trigger condition of compaction
+## for level-0 is that the number of the SST files reaches 4 (the default value). When both level-0
+## and level-1 adopt compaction, it is necessary to analyze RocksDB logs to know the size of an SST
+## file compressed from a MemTable. For example, if the file size is 32MB, the proposed value of
+## `max-bytes-for-level-base` is 32MB * 4 = 128MB.
 # max-bytes-for-level-base = "512MB"
 
-# Target file size for compaction.
+## Target file size for compaction.
+## The SST file size of level-0 is influenced by the compaction algorithm of `write-buffer-size`
+## and level0. `target-file-size-base` is used to control the size of a single SST file of level1 to
+## level6.
 # target-file-size-base = "8MB"
 
-# Max bytes for compaction.max_compaction_bytes
+## Max bytes for `compaction.max_compaction_bytes`.
+## If it's necessary to enlarge value of this entry, it's better to also enlarge `reserve-space`
+## in `storage` to ensure that a restarted TiKV instance can perform compactions successfully.
 # max-compaction-bytes = "2GB"
 
-# There are four different algorithms to pick files to compact.
-# 0 : ByCompensatedSize
-# 1 : OldestLargestSeqFirst
-# 2 : OldestSmallestSeqFirst
-# 3 : MinOverlappingRatio
+## There are four different compaction priorities.
+## 0 : ByCompensatedSize
+## 1 : OldestLargestSeqFirst
+## 2 : OldestSmallestSeqFirst
+## 3 : MinOverlappingRatio
 # compaction-pri = 3
 
-# block-cache used to cache uncompressed blocks, big block-cache can speed up read.
-# in normal cases should tune to 30%-50% system's total memory.
-# block-cache-size = "1GB"
-
-# Indicating if we'd put index/filter blocks to the block cache.
-# If not specified, each "table reader" object will pre-load index/filter block
-# during table initialization.
+## Indicating if we'd put index/filter blocks to the block cache.
+## If not specified, each "table reader" object will pre-load index/filter block during table
+## initialization.
 # cache-index-and-filter-blocks = true
 
-# Pin level0 filter and index blocks in cache.
+## Pin level-0 filter and index blocks in cache.
 # pin-l0-filter-and-index-blocks = true
 
-# Enable read amplication statistics.
-# value  =>  memory usage (percentage of loaded blocks memory)
-# 1      =>  12.50 %
-# 2      =>  06.25 %
-# 4      =>  03.12 %
-# 8      =>  01.56 %
-# 16     =>  00.78 %
+## Enable read amplification statistics.
+## value  =>  memory usage (percentage of loaded blocks memory)
+## 1      =>  12.50 %
+## 2      =>  06.25 %
+## 4      =>  03.12 %
+## 8      =>  01.56 %
+## 16     =>  00.78 %
 # read-amp-bytes-per-bit = 0
 
-# Pick target size of each level dynamically.
+## Pick target size of each level dynamically.
 # dynamic-level-bytes = true
 
-# Options for Column Family write
-# Column Family write used to store commit informations in MVCC model
+## Optimizes bloom filters. If true, RocksDB won't create bloom filters for the max level of
+## the LSM to reduce metadata that should fit in RAM.
+## This value is setted to true for `default` cf by default because its kv data could be determined
+## whether really exists by upper logic instead of bloom filters. But we suggest to set it to false
+## while using `Raw` mode.
+# optimize-filters-for-hits = true
+
+## Enable compaction guard, which is an optimization to split SST files at TiKV region boundaries.
+## The optimization can help reduce compaction IO, and allow us to use larger SST file size
+## (thus less SST files overall) while making sure we can still efficiently cleanup stale data on
+## region migration.
+## This config is available to default CF and write CF.
+# enable-compaction-guard = true
+
+## The lower bound of SST file size when compaction guard is enabled. The config prevent SST files
+## being too small when compaction guard is enabled.
+# compaction-guard-min-output-file-size = "8M"
+
+## The upper bound of SST file size when compaction guard is enabled. The config prevent SST files
+## being too large when compaction guard is enabled. This config overrides target-file-size-base
+## for the same CF.
+# compaction-guard-max-output-file-size = "128M"
+
+## Options for "Default" Column Family for `Titan`.
+[rocksdb.defaultcf.titan]
+## The smallest value to store in blob files. Value smaller than
+## this threshold will be inlined in base DB.
+## default: 1KB
+# min-blob-size = "1KB"
+
+## The compression algorithm used to compress data in blob files.
+## Compression method.
+##   no:     kNoCompression
+##   snappy: kSnappyCompression
+##   zlib:   kZlibCompression
+##   bzip2:  kBZip2Compression
+##   lz4:    kLZ4Compression
+##   lz4hc:  kLZ4HCCompression
+##   zstd:   kZSTD
+# default: lz4
+# blob-file-compression = "lz4"
+
+## Specifics cache size for blob records
+# default: 0
+# blob-cache-size = "0GB"
+
+## If the ratio of discardable size of a blob file is larger than
+## this threshold, the blob file will be GCed out.
+# default: 0.5
+# discardable-ratio = 0.5
+
+## The mode used to process blob files. In read-only mode Titan
+## stops writing value into blob log. In fallback mode Titan
+## converts blob index into real value on flush and compaction.
+## This option is especially useful for downgrading Titan.
+##   default:   kNormal
+##   read-only: kReadOnly
+##   fallback:  kFallback
+# default: normal
+# blob-run-mode = "normal"
+
+## If set true, values in blob file will be merged to a new blob file while
+## their corresponding keys are compacted to last two level in LSM-Tree.
+##
+## With this feature enabled, Titan could get better scan performance, and
+## better write performance during GC, but will suffer around 1.1 space
+## amplification and 3 more write amplification if no GC needed (eg. uniformly
+## distributed keys) under default rocksdb setting.
+##
+## Requirement: level_compaction_dynamic_level_base = true
+## default: false
+# level_merge = false
+
+## Use merge operator to rewrite GC blob index.
+## default: false
+# gc-merge-rewrite = false
+
+## Options for "Write" Column Family, which stores MVCC commit information
 [rocksdb.writecf]
+## Recommend to set it the same as `rocksdb.defaultcf.compression-per-level`.
 # compression-per-level = ["no", "no", "lz4", "lz4", "lz4", "zstd", "zstd"]
 # block-size = "64KB"
+
+## Recommend to set it the same as `rocksdb.defaultcf.write-buffer-size`.
 # write-buffer-size = "128MB"
 # max-write-buffer-number = 5
 # min-write-buffer-number-to-merge = 1
+
+## Recommend to set it the same as `rocksdb.defaultcf.max-bytes-for-level-base`.
 # max-bytes-for-level-base = "512MB"
 # target-file-size-base = "8MB"
 
-# in normal cases should tune to 10%-30% system's total memory.
-# block-cache-size = "256MB"
 # level0-file-num-compaction-trigger = 4
 # level0-slowdown-writes-trigger = 20
 # level0-stop-writes-trigger = 36
@@ -443,16 +751,19 @@ endpoints = ["127.0.0.1:2379"]
 # compaction-pri = 3
 # read-amp-bytes-per-bit = 0
 # dynamic-level-bytes = true
+# optimize-filters-for-hits = false
+# enable-compaction-guard = true
+# compaction-guard-min-output-file-size = "8M"
+# compaction-guard-max-output-file-size = "128M"
 
 [rocksdb.lockcf]
 # compression-per-level = ["no", "no", "no", "no", "no", "no", "no"]
 # block-size = "16KB"
-# write-buffer-size = "128MB"
+# write-buffer-size = "32MB"
 # max-write-buffer-number = 5
 # min-write-buffer-number-to-merge = 1
 # max-bytes-for-level-base = "128MB"
 # target-file-size-base = "8MB"
-# block-cache-size = "256MB"
 # level0-file-num-compaction-trigger = 1
 # level0-slowdown-writes-trigger = 20
 # level0-stop-writes-trigger = 36
@@ -461,9 +772,12 @@ endpoints = ["127.0.0.1:2379"]
 # compaction-pri = 0
 # read-amp-bytes-per-bit = 0
 # dynamic-level-bytes = true
+# optimize-filters-for-hits = false
+# enable-compaction-guard = false
 
 [raftdb]
-# max-sub-compactions = 1
+# max-background-jobs = 4
+# max-sub-compactions = 2
 # max-open-files = 40960
 # max-manifest-file-size = "20MB"
 # create-if-missing = true
@@ -480,21 +794,26 @@ endpoints = ["127.0.0.1:2379"]
 # wal-bytes-per-sync = "512KB"
 
 # info-log-max-size = "1GB"
-# info-log-roll-time = "0"
+# info-log-roll-time = "0s"
 # info-log-keep-log-file-num = 10
 # info-log-dir = ""
+# info-log-level = "info"
+# optimize-filters-for-hits = true
 
 [raftdb.defaultcf]
+## Recommend to set it the same as `rocksdb.defaultcf.compression-per-level`.
 # compression-per-level = ["no", "no", "lz4", "lz4", "lz4", "zstd", "zstd"]
 # block-size = "64KB"
+
+## Recommend to set it the same as `rocksdb.defaultcf.write-buffer-size`.
 # write-buffer-size = "128MB"
 # max-write-buffer-number = 5
 # min-write-buffer-number-to-merge = 1
+
+## Recommend to set it the same as `rocksdb.defaultcf.max-bytes-for-level-base`.
 # max-bytes-for-level-base = "512MB"
 # target-file-size-base = "8MB"
 
-# should tune to 256MB~2GB.
-# block-cache-size = "256MB"
 # level0-file-num-compaction-trigger = 4
 # level0-slowdown-writes-trigger = 20
 # level0-stop-writes-trigger = 36
@@ -503,15 +822,141 @@ endpoints = ["127.0.0.1:2379"]
 # compaction-pri = 0
 # read-amp-bytes-per-bit = 0
 # dynamic-level-bytes = true
+# optimize-filters-for-hits = true
+# enable-compaction-guard = false
+
+[raft-engine]
+## Enable raft-engine will ignore all settings about `raftdb`.
+# enable = false
+## Recovery mode. Candidates are `tolerate-corrupted-tail-records` and `absolute-consistency`.
+# recovery_mode = "tolerate-corrupted-tail-records"
+# bytes-per-sync = "256KB"
+# target-file-size = "128MB"
+# purge-threshold = "10GB"
+## Raft engine has builtin entry cache. `cache-limit` limits the memory usage of the cache.
+# cache-limit = "1GB"
 
 [security]
-# set the path for certificates. Empty string means disabling secure connectoins.
+## The path for TLS certificates. Empty string means disabling secure connections.
 # ca-path = ""
 # cert-path = ""
 # key-path = ""
+# cert-allowed-cn = []
+#
+## Avoid outputing data (e.g. user keys) to info log. It currently does not avoid printing
+## user data altogether, but greatly reduce those logs.
+## Default is false.
+# redact-info-log = false
+
+# Configurations for encryption at rest. Experimental.
+[security.encryption]
+## Encryption method to use for data files.
+## Possible values are "plaintext", "aes128-ctr", "aes192-ctr" and "aes256-ctr". Value other than
+## "plaintext" means encryption is enabled, in which case master key must be specified.
+# data-encryption-method = "plaintext"
+
+## Specifies how often TiKV rotates data encryption key.
+# data-key-rotation-period = "7d"
+
+## Enable an optimization to reduce IO and mutex contention for encryption metadata management.
+## Once the option is turned on (which is the default after 4.0.9), the data format is not
+## compatible with TiKV <= 4.0.8. In order to downgrade to TiKV <= 4.0.8, one can turn off this
+## option and restart TiKV, after which TiKV will convert the data format to be compatible with
+## previous versions.
+# enable-file-dictionary-log = true
+
+## Specifies master key if encryption is enabled. There are three types of master key:
+##
+##   * "plaintext":
+##
+##     Plaintext as master key means no master key is given and only applicable when
+##     encryption is not enabled, i.e. data-encryption-method = "plaintext". This type doesn't
+##     have sub-config items. Example:
+##     
+##     [security.encryption.master-key]
+##     type = "plaintext"
+##
+##   * "kms":
+##
+##     Use a KMS service to supply master key. Currently only AWS KMS is supported. This type of
+##     master key is recommended for production use. Example:
+##
+##     [security.encryption.master-key]
+##     type = "kms"
+##     ## KMS CMK key id. Must be a valid KMS CMK where the TiKV process has access to.
+##     ## In production is recommended to grant access of the CMK to TiKV using IAM.
+##     key-id = "1234abcd-12ab-34cd-56ef-1234567890ab"
+##     ## AWS region of the KMS CMK.
+##     region = "us-west-2"
+##     ## (Optional) AWS KMS service endpoint. Only required when non-default KMS endpoint is
+##     ## desired.
+##     endpoint = "https://kms.us-west-2.amazonaws.com"
+##
+##   * "file":
+##
+##     Supply a custom encryption key stored in a file. It is recommended NOT to use in production,
+##     as it breaks the purpose of encryption at rest, unless the file is stored in tempfs.
+##     The file must contain a 256-bits (32 bytes, regardless of key length implied by 
+##     data-encryption-method) key encoded as hex string and end with newline ("\n"). Example:
+##
+##     [security.encryption.master-key]
+##     type = "file"
+##     path = "/path/to/master/key/file"
+##
+# [security.encryption.master-key]
+# type = "plaintext"
+
+## Specifies the old master key when rotating master key. Same config format as master-key.
+## The key is only access once during TiKV startup, after that TiKV do not need access to the key.
+## And it is okay to leave the stale previous-master-key config after master key rotation.
+# [security.encryption.previous-master-key]
+# type = "plaintext"
 
 [import]
-# number of threads to handle RPC requests.
+## Number of threads to handle RPC requests.
 # num-threads = 8
-# stream channel window size, stream will be blocked on channel full.
+
+## Stream channel window size, stream will be blocked on channel full.
 # stream-channel-window = 128
+
+[backup]
+## Number of threads to perform backup tasks.
+## The default value is set to min(CPU_NUM * 0.75, 32).
+# num-threads = 24
+
+## Number of ranges to backup in one batch.
+# batch = 8
+
+## When Backup region [a,e) size exceeds `sst-max-size`, it will be backuped into several Files [a,b),
+## [b,c), [c,d), [d,e) and the size of [a,b), [b,c), [c,d) will be `sst-max-size` (or a
+## little larger).
+# sst-max-size = "144MB"
+
+[pessimistic-txn]
+## The default and maximum delay before responding to TiDB when pessimistic
+## transactions encounter locks
+# wait-for-lock-timeout = "1s"
+
+## If more than one transaction is waiting for the same lock, only the one with smallest
+## start timestamp will be waked up immediately when the lock is released. Others will
+## be waked up after `wake_up_delay_duration` to reduce contention and make the oldest
+## one more likely acquires the lock.
+# wake-up-delay-duration = "20ms"
+
+## Enable pipelined pessimistic lock, only effect when processing perssimistic transactions
+## Enabled this will improve performance, but slightly increase the transcation failure rate
+# pipelined = true
+
+[gc]
+## The number of keys to GC in one batch.
+# batch-keys = 512
+
+## Max bytes that GC worker can write to rocksdb in one second.
+## If it is set to 0, there is no limit.
+# max-write-bytes-per-sec = "0"
+
+## Enable GC by compaction filter or not.
+# enable-compaction-filter = true
+
+## Garbage ratio threshold to trigger a GC.
+# ratio-threshold = 1.1