[WIP][feat]: New configure.

.clang-tidy

@@ -135,6 +135,8 @@ Checks: "*,
   -readability-uppercase-literal-suffix,
   -readability-use-anyofallof,
   -readability-redundant-access-specifiers,
+  -readability-identifier-naming,
+  -readability-redundant-string-init,
 
   -zirkon-*,
   "

.gitignore

@@ -86,3 +86,4 @@ docker/rocky9/dingofs
 
 .cache
 dist/
+.playground

CMakeLists.txt

@@ -157,6 +157,9 @@ message("OpenSSL libraries: ${OPENSSL_LIBRARIES}, include dir: ${OPENSSL_INCLUDE
 find_package(Snappy)
 message("Using Snappy ${Snappy_VERSION}")
 
+find_package(toml11 REQUIRED)
+message("Using toml11 ${toml11_VERSION}")
+
 find_package(fmt REQUIRED)
 message("Using fmt ${fmt_VERSION}")
 

conf/client copy.conf

@@ -0,0 +1,202 @@
+##### mdsOpt
+# RPC total retry time with MDS
+mdsOpt.mdsMaxRetryMS=16000
+# The maximum timeout of RPC communicating with MDS.
+# The timeout of exponential backoff cannot exceed this value
+mdsOpt.rpcRetryOpt.maxRPCTimeoutMS=2000
+# RPC timeout for once communication with MDS
+mdsOpt.rpcRetryOpt.rpcTimeoutMs=500
+# RPC with mds needs to sleep for a period of time before each retry
+mdsOpt.rpcRetryOpt.rpcRetryIntervalUS=50000
+# Switch if the number of consecutive retries on the current MDS exceeds the limit.
+# The number of failures includes the number of timeout retries
+mdsOpt.rpcRetryOpt.maxFailedTimesBeforeChangeAddr=2
+# The normal retry times for trigger wait strategy
+mdsOpt.rpcRetryOpt.normalRetryTimesBeforeTriggerWait=3
+# Sleep interval for wait
+mdsOpt.rpcRetryOpt.waitSleepMs=1000
+mdsOpt.rpcRetryOpt.addrs=127.0.0.1:6700,127.0.0.1:6701,127.0.0.1:6702  # __ANSIBLE_TEMPLATE__ {{ groups.mds | join_peer(hostvars, "mds_listen_port") }} __ANSIBLE_TEMPLATE__
+
+
+#
+# lease options
+#
+# client and mds lease time, default is 20s
+mds.leaseTimesUs=20000000
+mds.refreshTimesPerLease=5
+
+#### metaCacheOpt
+# Gets the number of retries for the leader
+metaCacheOpt.metacacheGetLeaderRetry=3
+# Need to sleep for a period of time before each get leader retry
+metaCacheOpt.metacacheRPCRetryIntervalUS=100000
+# RPC timeout of get leader
+metaCacheOpt.metacacheGetLeaderRPCTimeOutMS=1000
+
+#### executorOpt
+# executorOpt rpc with metaserver
+# rpc retry times with metaserver
+executorOpt.maxRetry=10
+# internal rpc retry times with metaserver
+executorOpt.maxInternalRetry=3
+# Retry sleep time between failed RPCs
+executorOpt.retryIntervalUS=100000
+# RPC timeout for communicating with metaserver
+executorOpt.rpcTimeoutMS=1000
+# RPC stream idle timeout
+executorOpt.rpcStreamIdleTimeoutMS=500
+# The maximum timeout RPC time of the retry request.
+# The timeout time will follow the exponential backoff policy.
+# Because the timeout occurs when the network is congested, the RPC timeout needs to be increased
+executorOpt.maxRPCTimeoutMS=8000
+# Maximum sleep time between retry requests.
+# when the network is congested or the metaserver is overloaded,
+# it is necessary to increase the sleep time
+executorOpt.maxRetrySleepIntervalUS=8000000
+executorOpt.minRetryTimesForceTimeoutBackoff=5
+executorOpt.maxRetryTimesBeforeConsiderSuspend=20
+# batch limit of get inode attr and xattr
+executorOpt.batchInodeAttrLimit=10000
+
+#### brpc
+# close socket after defer.close.second
+rpc.defer.close.second=1
+# rpc health check interval in second, 0 or negative value means disable health check
+rpc.healthCheckIntervalSec=0
+
+#### fuseClient
+# TODO(xuchaojie): add unit
+fuseClient.listDentryLimit=65536
+fuseClient.downloadMaxRetryTimes=3
+
+
+# you shoudle enable it when mount one filesystem to multi mountpoints,
+# it gurantee the consistent of file after rename, otherwise you should
+# disable it for performance.
+fuseClient.enableMultiMountPointRename=true
+# thread number of listDentry when get summary xattr
+fuseClient.listDentryThreads=10
+# default data（s3ChunkInfo/volumeExtent） size in inode, if exceed will eliminate and try to get the merged one
+fuseClient.maxDataSize=1024
+# default refresh data interval 30s
+fuseClient.refreshDataIntervalSec=30
+fuseClient.warmupThreadsNum=10
+
+# when read happen all block of file will be prefetch to block cache
+# if no disk cache this option will be ignored
+fuseClient.in_time_warmup=false
+
+fuseClient.bthread_worker_num=0
+
+
+fs.rpc.listDentryLimit=65536
+fs.deferSync.delay=3
+fs.deferSync.deferDirMtime=false
+# }
+
+
+
+#### block cache
+# {
+# block_cache.cache_store:
+#   cache store type, none, disk or 3fs
+#
+# block_cache.stage_bandwidth_throttle_enable:
+#   block will been put to s3 storage directly if disk write bandwidth
+#   exceed limit.
+#
+# disk_cache.cache_dir:
+#   directory for store cache block, multi directories
+#   and corresponding max size are supported, e.g. "/data1:200;/data2:300"
+#
+# disk_cache.ioring_iodepth:
+#   iodepth for io ring (works for both linux io uring and 3fs usrbio)
+block_cache.cache_store=disk
+block_cache.stage=true
+block_cache.stage_bandwidth_throttle_enable=false
+block_cache.stage_bandwidth_throttle_mb=10240
+block_cache.logging=true
+block_cache.upload_stage_workers=10
+block_cache.upload_stage_queue_size=10000
+block_cache.prefetch_workers=128
+block_cache.prefetch_queue_size=10000
+
+disk_cache.cache_dir=/var/run/dingofs  # __DINGOADM_TEMPLATE__ /dingofs/client/data/cache __DINGOADM_TEMPLATE__
+disk_cache.cache_size_mb=102400
+disk_cache.free_space_ratio=0.1
+disk_cache.cache_expire_second=259200
+disk_cache.cleanup_expire_interval_millsecond=1000
+disk_cache.drop_page_cache=true
+disk_cache.ioring_iodepth=128
+disk_cache.ioring_blksize=1048576
+disk_cache.ioring_prefetch=true
+
+disk_state.tick_duration_second=60
+disk_state.normal2unstable_io_error_num=3
+disk_state.unstable2normal_io_succ_num=10
+disk_state.unstable2down_second=1800
+disk_state.disk_check_duration_millsecond=3000
+# }
+
+#### s3
+# this is for test. if s3.fakeS3=true, all data will be discarded
+s3.fakeS3=false
+# prefetch blocks that disk cache use
+s3.prefetchBlocks=1
+# prefetch threads
+s3.prefetchExecQueueNum=1
+# start sleep when mem cache use ratio is greater than nearfullRatio,
+# sleep time increase follow with mem cache use ratio, baseSleepUs is baseline.
+s3.nearfullRatio=70
+s3.baseSleepUs=500
+
+# write cache < 8,388,608 (8MB) is not allowed
+s3.writeCacheMaxByte=838860800
+s3.readCacheMaxByte=209715200
+# file cache read thread num
+s3.readCacheThreads=5
+
+s3.verify_SSL=False
+s3.region=us-east-1
+s3.maxConnections=32
+s3.connectTimeout=60000
+s3.requestTimeout=10000
+# Off = 0,Fatal = 1,Error = 2,Warn = 3,Info = 4,Debug = 5,Trace = 6
+s3.logLevel=4
+s3.logPrefix=/data/logs/dingofs/aws_ # __DINGOADM_TEMPLATE__ /dingofs/client/logs/aws_ __DINGOADM_TEMPLATE__
+# limit all inflight async requests' bytes, |0| means not limited
+s3.maxAsyncRequestInflightBytes=104857600
+# throttle
+s3.throttle.iopsTotalLimit=0
+s3.throttle.iopsReadLimit=0
+s3.throttle.iopsWriteLimit=0
+s3.throttle.bpsTotalMB=0
+s3.throttle.bpsReadMB=0
+s3.throttle.bpsWriteMB=0
+s3.useVirtualAddressing=false
+# The interval between read failures and retries will become larger and larger,
+# and when the max is reached, retry will be performed at a fixed time.
+s3.maxReadRetryIntervalMs = 1000
+# retry interval
+s3.readRetryIntervalMs = 100
+
+s3.enableTelemetry=false
+
+s3.use_crt_client=false
+# this only work when use_crt_client is false
+s3.use_thread_pool=true
+# this only work when use_crt_client is false and use_thread_pool is true
+s3.async_thread_num_in_thread_pool=256
+
+#### common
+client.common.logDir=/data/logs/dingofs  # __DINGOADM_TEMPLATE__ /dingofs/client/logs __DINGOADM_TEMPLATE__
+# we have loglevel: {0,3,6,9}
+# as the number increases, it becomes more and more detailed
+client.loglevel=0
+client.dummyServer.startPort=9000
+
+
+mds.addr=172.20.61.102:7801
+
+### uds
+uds.fdCommPath=/var/run # unix domain socket file path

conf/client.conf

@@ -280,7 +280,7 @@ s3.enableTelemetry=false
 
 s3.use_crt_client=false
 # this only work when use_crt_client is false
-s3.use_thread_pool=true 
+s3.use_thread_pool=true
 # this only work when use_crt_client is false and use_thread_pool is true
 s3.async_thread_num_in_thread_pool=256
 

confv2/dingo-cache.gflags

@@ -0,0 +1,11 @@
+-block_cache_logging=true
+-block_cache_stage_bandwidth_throttle_enable=false
+-block_cache_stage_bandwidth_throttle_mb=10240
+-disk_cache_cache_expire_s=259200
+-disk_cache_cleanup_expire_interval_ms=1000
+-disk_cache_drop_page_cache=true
+-disk_state_tick_duration_s=60
+-disk_state_normal2unstable_io_error_num=3
+-disk_state_unstable2normal_io_succ_num=10
+-disk_state_unstable2down_s=1800
+-disk_state_disk_check_duration_ms=3000

confv2/dingo-cache.template.toml

@@ -0,0 +1,37 @@
+[cache_group]
+group_name = "default"
+listen_ip = "127.0.0.1"
+listen_port = 9302
+group_weight = 100
+max_range_size_kb = 256
+metadata_filepath = "/var/log/cache_group_meta"
+load_members_interval_ms = 1000
+
+[block_cache]
+logging = true
+cache_store = "disk"
+stage = true
+stage_bandwidth_throttle_enable = false
+stage_bandwidth_throttle_mb = 10240
+upload_stage_workers = 10
+upload_stage_queue_size = 10000
+prefetch_workers = 128
+prefetch_queue_size = 10000
+
+[block_cache.disk_cache]
+cache_dir = [ '/var/run/dingofs' ]
+cache_size_mb = 10240
+free_space_ratio = 0.1
+cache_expire_s = 259200
+cleanup_expire_interval_ms = 1000
+drop_page_cache = true
+ioring_iodepth = 128
+ioring_blksize = 1048576
+ioring_prefetch = true
+
+[block_cache.disk_cache.disk_state]
+tick_duration_s = 60
+normal2unstable_io_error_num = 3
+unstable2normal_io_succ_num = 10
+unstable2down_s = 1800
+disk_check_duration_ms = 3000

confv2/dingo-fuse.gflags

@@ -0,0 +1 @@
+-access_lgging=true