It's my first time using RDS on AWS, I use a t2.medium instance running MySQL Aurora with default configs. The CPU usage and DB connections is quite normal until "something" happen, which cause the DB connections go all the way up to it's maximum (with t2.medium it is 80 connections).

enter image description here

I only have one web app, run on an EC2 instance, when the DB connections reach it's maximum, the EC2 instance's CPU usage is absolutely normal (25-30%), all attempts to connect to DB instance result in "Too many connections".

enter image description here

I also check the DB instance's CPU Utilization at the time, it was showing no signal of high load. During the strike time, CPU Utilization dropped to 20% and keep that rate consistently.

enter image description here

Thing I don't understand: DB connections is at maximum, but why DB CPU Utilization is dropped? Shouldn't it be maximum as well because of the computation for the queries in those connections?

Please help me understand, thank you very much.

(I had to resize the RDS instance to r4.large when the second strike happens, still running it now until I find out the problem...)

  • Is the slowlog turned on? What was the I/O like during the spike? – Rick James May 29 '18 at 23:21
  • Additional information request. Post on pastebin.com and share the links. Current RAM size of your MySQL Host server A) complete (not edited) my.cnf or my.ini Text results of: B) SHOW GLOBAL STATUS; after minimum 24 hours UPTIME C) SHOW GLOBAL VARIABLES; D) SHOW FULL PROCESSLIST; AND Optional very helpful information, if available includes - htop OR top OR mytop for most active apps, ulimit -a for a linux/unix list of limits, iostat -xm 5 3 for IOPS by device and core/cpu count, df -h for a linux/unix free space list by device, for server tuning analysis. – Wilson Hauck yesterday
  • Please post results of SHOW GLOBAL STATUS LIKE 'threads%'; and SHOW GLOBAL VARIABLES LIKE 'thread%'; – Wilson Hauck 12 hours ago

If you run show processlist; it'll show you all the connections that are running against your DB. Running show status like 'Conn%' can show how many connections are active

You may think that a high number of connections would increase the CPU but ultimately it depends what those connections are trying to accomplish, if they aren't doing much but are preventing you're main application from running their connections then the normal processing level will go down while you are not running your normal processes

The annoying thing is to be able to run the show processlist command you need to actually be connected to the DB which you wont be able to as you're at max connections.

A slight work around is to increase your max connections (if you're able to) and set up some monitoring somewhere that runs every so often (every 5 minutes would have caught that) and when you're over 50 connections, run a command that dumps out all the active connections somewhere so you can review later.


In your RDS, first check all IOPS metrics.

RDS provide a certain quantity of IOPS, and if your RDS Instance have 10GB SSD, probably your IOPS quantity is 100 IOPS (default), this value increase dependends of the instance SSD size.

If you have not IOPS enough, your RDS is blocked for IO operations, causing 'Sending Data' in your MySQL and high timeouts, but no change CPU usage or Memory, because this is an Disk issue.

For fix it, change the configurations of the MySQL, in Dashboard you change it.

I have MySQL on EC2 instance, low cost for me.

See my.cnf


port                           = 3306
user                           = mysql
default-storage-engine         = InnoDB
socket                         = /var/lib/mysql/mysql.sock
pid-file                       = /var/lib/mysql/mysql.pid

log-error                      = /var/lib/mysql/mysql-error.log
log-queries-not-using-indexes  = 1
slow-query-log                 = 0
slow-query-log-file            = /var/lib/mysql/mysql-slow.log
log_error_verbosity            = 2

max-allowed-packet              = 6M
sysdate-is-now                  = 1

datadir                         = /var/lib/mysql

key-buffer-size                 = 128M
query_cache_size                = 128M #100M384M

tmp_table_size                  = 256M
max_heap_table_size             = 256M

innodb-buffer-pool-size         = 2G 
innodb_log_buffer_size          = 8M
innodb_log_file_size            = 1G

wait_timeout                    = 10
interactive_timeout             = 300

max-connect-errors              = 100000
max-connections                 = 200

sort_buffer_size                = 4M
read_buffer_size                = 2M
read_rnd_buffer_size            = 2M
join_buffer_size                = 2M
thread_stack                    = 4M
thread-cache-size               = 80

performance_schema              = on

query_cache_type                = 1 #0 #1
query_cache_limit               = 128M

table_open_cache                = 2680
open-files-limit                = 1024000
table-definition-cache          = 3024


innodb-flush-method            = O_DIRECT
innodb-log-files-in-group      = 2
innodb-flush-log-at-trx-commit = 2

innodb_buffer_pool_instances   = 1
innodb_stats_on_metadata       = 0

innodb_io_capacity             = 100
innodb_use_native_aio          = 1
innodb-file-per-table          = 0

explicit_defaults_for_timestamp = 1

[This conf is for 2 vcpu and 4GB RAM, 10GB SSD with 100IOPS] Work fine for me in AWS NLB + Auto Scaling Service.

Other optimizations is would be increased the innodb buffer to 50% of memory of the instance, and innodb_log_file_size with 50% of the innodb-buffer-pool-size value.

Read this: https://dev.mysql.com/doc/refman/5.5/en/optimizing-innodb-diskio.html

I hope help you or other users with same problem. =)

New contributor
Gustavo alves is a new contributor to this site. Take care in asking for clarification, commenting, and answering. Check out our Code of Conduct.
  • If you would put # and a space character in front of your read_rnd_buffer_size you would find your RPS of handler_read_rnd_next would be much lower. For other suggestions, view my profile, Network profile for contact information and get in touch with Skype, please. – Wilson Hauck yesterday

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.