问题
After upgrading my RDS to a larger instance class, I started seeing packet sequence number wrong in my error logs.
I have 2 web servers created by the ElasticBeanstalk.
How do I got rid of the error?
Most of the literature (here and on the web in general) blame this error on multithreading, which we're not using.
回答1:
The solution was to reboot the instances, one by one.
The safest way seemed to be from the Health tab of the ElasticBeanstalk; select each instance, and under Instance Action there's a reboot option.
After the reboots, the packet sequence number wrong error was never seen again.
It seems that the 75 seconds downtime was not long enough for the code to realize it had to (close the old and) open new database connections, and was using non-existant handles.
来源：https://stackoverflow.com/questions/46770023/error-packet-sequence-number-wrong-after-upgrading-rds-instance