Re: MySQL over TCP results on CLOSE_WAIT state in PHP 5.3.8

> > I have a MySQL server A, a server B with PHP 5.3.8 and a server C with > > PHP 5.3.3. I'm connecting to the MySQL server on A via TCP from B and C > > using an internal network. Server B and C use the same PHP application. > > There are also same PHP scripts that get data from the database and then > > calculate up to 30 minutes. I close all database connection before doing > > the calculation to save connections (and ports) using: > > $thread_id = mysqli_thread_id( $this->handle ); > > mysqli_kill( $this->handle, $thread_id ); > > mysqli_close( $this->handle ); > > > > During a review on our servers I discovered that server B has a lot of > > network connection in the state "CLOSE_WAIT". Server C running the same > > PHP application has not. I see the difference that server B is using > > mysqlnd and server C not. > > > > serverB# netstat -an | grep 3306 > > tcp 1 0 10.8.0.58:47455 10.8.0.1:3306 > > CLOSE_WAIT > > > > serverA# cat firewall > > Feb 17 16:21:49 www kernel: [6587053.325075] SFW2-OUT-ERROR IN= OUT=tun0 > > SRC=10.8.0.1 DST=10.8.0.58 LEN=40 TOS=0x00 PREC=0x00 TTL=64 ID=0 DF > > PROTO=TCP SPT=3306 DPT=47455 WINDOW=0 RES=0x00 RST URGP=0 > > > > Does anybody have an idea why this happens? How can I avoid this or > > investigate in this? Is this a know issue? > > Is the underlying OS for servers B & C exactly the same? Down to > kernel version (if Linux/Unix), NICs and driver version, > configurations, etc.? It would help if specify OS type and kernel > version. No, client B and C have different OS versions. You are right this can be a reason too. Do you have any hints how I can trace that down? Details client B (newer OS with CLOSE_WAIT problems): $ uname -a Linux www3 3.1.9-1.4-default #1 SMP Fri Jan 27 08:55:10 UTC 2012 (efb5ff4) x86_64 x86_64 x86_64 GNU/Linux $ mysql --version mysql Ver 14.14 Distrib 5.5.16, for Linux (x86_64) using readline 6.2 $ ethtool -i eth0 driver: r8169 version: 2.3LK-NAPI firmware-version: rtl_nic/rtl8168e-2.fw bus-info: 0000:04:00.0 supports-statistics: yes supports-test: no supports-eeprom-access: no supports-register-dump: yes Details client C: $ uname -a Linux www13 2.6.34.10-0.4-default #1 SMP 2011-10-19 22:16:41 +0200 x86_64 x86_64 x86_64 GNU/Linux $ mysql --version mysql Ver 14.14 Distrib 5.1.57, for suse-linux-gnu (x86_64) using readline 6.1 $ ethtool -i eth0 driver: r8169 version: 2.3LK-NAPI firmware-version: bus-info: 0000:04:00.0 Server A: Linux www 3.1.9-1.4-default #1 SMP Fri Jan 27 08:55:10 UTC 2012 (efb5ff4) x86_64 x86_64 x86_64 GNU/Linux $ mysql --version mysql Ver 14.14 Distrib 5.5.16, for Linux (x86_64) using readline 6.2 $ ethtool -i eth0 driver: r8169 version: 2.3LK-NAPI firmware-version: rtl_nic/rtl8168e-2.fw bus-info: 0000:04:00.0 supports-statistics: yes supports-test: no supports-eeprom-access: no supports-register-dump: yes I also have set wait_timeout 45 in my.cnf on server A and mysqli.reconnect On on client B and C in php.ini.