Search Postgresql Archives

max_connections reached in postgres 9.3.3

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Our set up:

·         Db version: postgres 9.3.3

·         OS: CentOS 6.5

·         kernel Version - Linux 2.6.32-431.3.1.el6.x86_64 #1 SMP Fri Jan 3 21:39:27 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux

·         cpu - 24 proc

·         memory - 768 GB

·         The disks are SAN fiber.

·         We have streaming replication set up from this server to the secondary server.

·         Some of the DB parameters:

max_connections  - 1500

shared_buffers - 4GB

work_mem = 130MB             

 maintenance_work_mem = 1GB

 

Issue this morning:

At 9:04 - FATAL:  remaining connection slots are reserved for non-replication superuser connections

These were the things we noticed and have question on each one of them:

pg_stat_activity showed a total of 850 connections but a 'ps -ef|grep postgers' showed 1500 connections and DB wasn’t accepting new connections. How is it possible to have such discrepancy in count between pg_stat_activity and ps?

200 were in waiting status, 50 were in authentication status, 748 were in IDLE status and the rest were all doing some DML operations. Why will 50 connections be in authentication status in the database and stay there?

 Some existing sessions were completing the sqls successfully (though not as fast). So, the database was not completely hung or unresponsive.

Trying to kill the session at the DB level (pg_terminate_backend) were just hanging. Why would this be the case?

During this time, server had a very high load average of 130 (normally we have < 5).

We had to restart the database to clear it all up.

 Note:  

We are having this issue of max_connections being reached for a couple of months now. At one point, we tried increasing the max_connections to 2500 to see if we will still run out of connections, and we did.

We noticed that generic-segmentation-offload was set to ON on this server. We changed it to OFF.

 

Recent Changes in the environment:

·         We turned off THP on the server a couple of months ago to see if that will fix the issue.

·         We recently (2 weeks ago) moved from RedHat ext 2 to xfs because we were having the same problem.

 

Please let us know if you need more information on our environment or on the issue this morning. Any help will be greatly appreciated.

 

Thank you,

________________

Ramya Vasudevan

Database Administrator

CLASSMATES

 


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Index of Archives]     [Postgresql Jobs]     [Postgresql Admin]     [Postgresql Performance]     [Linux Clusters]     [PHP Home]     [PHP on Windows]     [Kernel Newbies]     [PHP Classes]     [PHP Books]     [PHP Databases]     [Postgresql & PHP]     [Yosemite]
  Powered by Linux