GitHub user 903174293 closed a discussion: kvrocks During a 10-minute stress
test, the master node became unavailable and reported an error. E20240808
16:57:31.393215 14244 replication.cc:146] Write error while sending batch to
slave: Connection reset by peer. batches:
1、kvrocks version 2.7.0
Node specifications 16C32G
10 Sharding 3Replica
worker 8
Key parameters in the configuration file:
**_rocksdb.block_cache_size 4096
rocksdb.row_cache_size 0
rocksdb.max_open_files 8096
rocksdb.write_buffer_size 64
rocksdb.target_file_size_base 128
rocksdb.max_write_buffer_number 4
rocksdb.block_size 16384
max-replication-mb 300
max-io-mb 0
max-db-size 0_**
2、Use the following command to insert data.
_**nohup memtier_benchmark -s x.x.x.x -p 6666 -a xxxxxx --cluster-mode
--print-percentiles 50,90,95,99,100 --random-data --randomize
--distinct-client-seed --hide-histogram --key-minimum 1 --key-maximum 100000000
--key-prefix="type_string_001_" --command="set __key__ __data__"
--command-ratio=1 --command-key-pattern=S -n 10000000 -c 1 -t 1
--data-size-range 32-4096 > logs/${currentTime}/result-${currentTime}-1.log
2>&1 &**_
[createData.txt](https://github.com/user-attachments/files/16579290/createData.txt)
3、For about 10 minutes, some master nodes in the cluster were unavailable, and
the main node reported the following error:
E20240808 16:57:31.393215 14244
[replication.cc:146](http://replication.cc:146/)] Write error while sending
batch to slave: Connection reset by peer. batches:
E20240808 16:57:31.393215 14244
[replication.cc:146](http://replication.cc:146/)] Write error while sending
batch to slave: Connection reset by peer. batches:
No obvious abnormalities in the network, disk, CPU, or memory of the slave
nodes.
4、Phenomenon: The client cannot connect to the main node.
No response packet captured.
GitHub link: https://github.com/apache/kvrocks/discussions/2486
----
This is an automatically sent email for [email protected].
To unsubscribe, please send an email to: [email protected]
