This project is a command line tool as a controller for Apache Kvrocks. It also supports Redis cluster.
- meet node
- forget node
- import from exist kvrocks/redis cluster
- migrate slots, support given any slots from one node to another
- balance slots range
- tmp save migrated slots when node crash down and continue after restart
- failover
- check cluster nodes state/nodes number/slot range
docker run -it --name=segment_kvrocks_controller --net=host -v /opt/data:/opt/data key232323/segment_kvrocks_controller:latest bash
java -jar segment_kvrocks_controller-1.0.jar dbDataFile=/opt/data/kvrocks_controller_dataTIPS: This is for kvrocks cluster, set app.engine.isRedis=1 for redis cluster.
java -jar segment_kvrocks_controller-1.0.jar app.engine.isRedis=0TIPS: Your need install jdk8+/gradle6.x+
- mkdir /opt/your-dir
- cd /opt/your-dir
- git clone [email protected]:segment11/segment_kvrocks_controller.git segment_kvrocks_controller
- cd segment_kvrocks_controller
- gradle jar
- cd build/libs
- java -jar segment_kvrocks_controller-1.0.jar
This project has a default conf.properties in jar. You can overwrite configure item values when start.
java -jar segment_kvrocks_controller-1.0.jar app.engine.isRedis=0 dbDataFile=/your-dir/kvrocks_controller_dataTips: Item dbDataFile is a H2 database local file.
dbDataFile=D:/opt/kvrocks_controller_data
app.engine.isRedis=0
isLocalTest=0
job.migrate.wait.once.ms=10
# 2min
job.migrate.wait.max=12000
job.offset.check.wait.once.ms=100
# 5min
job.offset.check.wait.max=3000
# for redis
job.migrate.keys.once.number=100
jedis.pool.maxTotal=10
jedis.pool.maxIdle=5
jedis.pool.maxWait.ms=5000
jedis.read.timeout.ms=10000
#app.globalName=test
#app.globalPassword=test1234
#cli.runtime.jar=1
Usage:
usage: please input follow args to run task
-A,--check_all_shard_node check all shard nodes
cluster info/nodes/slots
if ok and match
-a,--meet_node operation: add one shard
node to cluster
-b,--init_batch <arg> operation: add some
primary shard nodes to
cluster from beginning,
eg.
-b=192.168.99.100:6379,19
2.168.99.100:6380
-C,--clear clear old application
saved local
-c,--check_shard_node check one shard node
cluster info/nodes/slots
if ok and match
-d,--forget_node operation: remove one
shard node from cluster
-D,--delete delete application local
-f,--failover operation: failover one
shard replica node as
primary
-F,--migrate_slots <arg> migrate some slots to
current session ip/port
shard node, eg. -F=0
-h,--ip <arg> target node ip
-H,--help args help
-I,--import <arg> import new application
from exist kvrocks
cluster, require ip:port,
eg.
-I=192.168.99.100:6379
-J,--job_log display job log
-L,--delete_job_log <arg> delete job log by id, eg.
-L=* or -L=1,2,3
-m,--is_primary is primary, eg. true or
false
-M,--fix_migrating_node fix refresh cluster nodes
after restart when
migrate job undone
@deprecated
-n,--name <arg> define a unique
application name for
target kvrocks cluster,
eg. myCluster1 or
testCluster
-N,--down_shard_node set target shard node
down
-O,--redo_one_job_by_id_force redo one job by one job
log id ignore done and
result is ok, eg. -R=1 -O
-p,--password <arg> kvrocks server password
-P,--port <arg> target node port
-Q,--quit quit console
-r,--replica_index <arg> target replica index, eg.
0 or 1
-R,--redo_one_job_by_id <arg> redo one job by one job
log id, eg. -R=1
-S,--shard_slot_range_re_avg <arg> migrate other slot to
target primary node so
this shard slot range
will be avg, require
ip:port, eg.
-S=192.168.99.100:6379
-s,--shard_index <arg> target shard index, eg. 0
or 1
-T,--delete_tmp_saved_migrated_slot_log <arg> delete by job log id, eg.
-T=* or -T=1,2,3
-U,--tmp_saved_migrated_slot_log display tmp saved
migrated slot log
-v,--view_shard_node view one shard node
cluster info/nodes/slots
-V,--view_shard_detail view all shard nodes
-X,--x_session_current_variables view args for reuse in
current session
-Y,--up_shard_node set target shard node up
-z,--lazy_migrate do not migrate slots
after adding one shard
node to cluster TIPS:
--name/--password/--ip/--port will be cached in the current session, so you can omit them.
Import from a cluster:
--import=192.168.99.100:6379 --name=TestCluster --password=***Add a primary shard node:
-a -m -s=0 --ip=192.168.99.100 --port=6379 --name=TestCluster --password=***
-a -m -s=1 --ip=192.168.99.100 --port=6380
-a -m -s=2 --ip=192.168.99.100 --port=6381Add some primary shard nodes from beginning:
-b=192.168.99.100:6379,192.168.99.100:6380Add a replica shard node:
-a -s=0 -r=0 --ip=192.168.99.100 --port=6479Delete one shard node:
-d --ip=192.168.99.100 --port=6479Migrate some slots to target shard node:
-F=0-100,100,200 --ip=192.168.99.100 --port=6380Balance all slots range avg to one shard primary node:
-S=192.168.99.100:6380Failover one shard to target replica shard node:
-f -s=0 -r=0Fix one shard node when crash down, after restart kvrocks process:
-M --ip=192.168.99.100 --port=6380Clear local H2 database records, then you can import or add shard node from beginning:
--clear -V
# or
-DView the current session variables:
-XView the current cluster all shard nodes slots range:
-V
# view only one shard node
-v --ip=192.168.99.100 --port=6380Check the current cluster all shard nodes state:
-A
# check only one shard node
-c --ip=192.168.99.100 --port=6380View the current cluster jobs:
-J
# redo one job
-R=jobId1
# redo one job force
-R=jobId1 -O
# delete target job records
-L=jobId1,jobId2
# delete all job records
-L=*
# view tmp saved migrated slots record list in history jobs (eg. see more detail for migrate slots using -F=0-10 --ip=** --port=**)
-U
# delete tmp saved migrated slots record in one job (usually for crash slot range total overlay situation)
-T=jobId1,jobId2
-T=*Others:
# help
-H
# quit
-Q
# set target shard node kvrocks process status running (usually for in maintain or test)
-Y --ip=192.168.99.100 --port=6380
# set target shard node kvrocks process status stopped (usually for in maintain or test)
-N --ip=192.168.99.100 --port=6380