mysql 内存持续增长，可以从哪里着手排查呢？

[cmai]

my.conf 如下 [mysqld]

character_set_server            = utf8mb4

datadir                         = /var/lib/mysql

expire_logs_days                = 1

explicit_defaults_for_timestamp = 1

general_log                     = ON

general_log_file                = /var/log/mysql/general_log_file.log

innodb_buffer_pool_size         = 1G

innodb_flush_log_at_trx_commit  = 2

innodb_flush_method             = O_DIRECT

innodb_flush_neighbors          = 0

innodb_io_capacity              = 1000

innodb_io_capacity_max          = 2000

innodb_large_prefix             = 1

innodb_lock_wait_timeout        = 30

innodb_print_all_deadlocks      = 1

innodb_thread_concurrency       = 4

join_buffer_size                = 1M

log-error                       = /var/log/mysql/error.log

log_queries_not_using_indexes   = 0

log_slow_admin_statements       = 1

log_slow_slave_statements       = 1

log_timestamps                  = system

long_query_time                 = 10

max_connect_errors              = 10

pid-file                        = /var/run/mysqld/mysqld.pid

read_rnd_buffer_size            = 8388608

slow_query_log                  = 1

slow_query_log_file             = /var/log/mysql/slow_query_log_file.log

socket                          = /var/run/mysqld/mysqld.sock

sort_buffer_size                = 4194304

tmp_table_size                  = 67108864

wait_timeout                    = 600

binlog-ignore-db                = mysql

enforce-gtid-consistency        = ON

gtid-mode                       = ON

log-bin                         = mysql-bin

log-slave-updates               = ON

innodb_log_file_size            = 256M

lower_case_table_names          = 1

max_connections                 = 512

server-id                       = 1

sql-mode                        =
STRICT_TRANS_TABLES,NO_ZERO_IN_DATE,NO_ZERO_DATE,ERROR_FOR_DIVISION_BY_ZERO,NO_AUTO_CREATE_USER,NO_ENGINE_SUBSTITUTION

[cmai]

error log 里有频繁的这个信息

2023-03-27T14:25:34.279971+08:00 41869 [Note] Got an error reading communication packets
2023-03-27T14:25:54.280650+08:00 41876 [Note] Got an error reading communication packets
2023-03-27T14:25:54.280710+08:00 41877 [Note] Got an error reading communication packets
2023-03-27T14:26:14.280804+08:00 41881 [Note] Got an error reading communication packets
2023-03-27T14:26:14.280832+08:00 41882 [Note] Got an error reading communication packets
2023-03-27T14:26:34.280017+08:00 41885 [Note] Got an error reading communication packets
2023-03-27T14:26:34.280043+08:00 41886 [Note] Got an error reading communication packets
2023-03-27T14:26:54.280510+08:00 41891 [Note] Got an error reading communication packets
2023-03-27T14:26:54.280538+08:00 41890 [Note] Got an error reading communication packets

[cmai]

使用 k8s 部署的， 内存增长到 limit 上限就重启了

[cmai]

@cmai 以及有大量的 2023-03-27T14:28:59.264149+08:00 41757 [Note] Aborted connection 41757

[jeffrey921]

有可能是查询数据后没释放掉

[cmai]

@jeffrey921 有办法验证吗

[sunjiayao]

先 show processlist 看看

[liprais]

日志挂出来了?

[CharAct3]

可以设置一下 maxLifetime ，定期重建连接，触发内存回收，可能对你们的 case 会有帮助

[Cornstalk8256]

多大的业务啊，MySQL 居然部署在容器里面

[cmai]

@sunjiayao 额，这个的目的是什么， 只有 74 个连接，大部分在 sleep 状态

[cmai]

统一回复，大家不用纠结容器的问题，我们做的是 devops 平台，目的在于一键部署一个 mysql ，并且有统一的指标监控、告警、备份策略以及配置等， 在于开发提效和分离开发人员的关注度

[cmai]

@CharAct3 ok ，我先了解下

[a7851578]

找个实例挂上 valgrind 看看，固定大小不能访问内存，可能是内存泄漏了