Blogs

Below one liner, pipes mailq command to sed, which prints out every 3rd line starting from line 5 capturing only the email addresses and producing report of the top 10 email addresses in sendmail mail queue.

"wait_timeout" is the amount of seconds during inactivity that MySQL will wait before it will close a connection on a non-interactive connection.

"interactive_timeout" is the same, but for interactive mysql shell sessions.

Setting a value too low may cause connections to drop unexpectedly, specifically if you are using persistent connections. Setting a value too high may cause stale connections to remain open, preventing new access to the database.

For wait_timeout, this value should be set as low as possible without affecting availability and performance.

For interactive_timeout, changing this value won't really increase or decrease performance of your application.

While install the puppet dashboard, I noticed puppetlabs yum repo changed from using "http://yum.puppetlabs.com/RPM-GPG-KEY-reductive" to "http://yum.puppetlabs.com/RPM-GPG-KEY-puppet" causing yum to fail on the key import. I had run the key import manually couple times before I realized, yum was complaining about a different key.

Below is how I got around to removing all of the imported gpgkeys:

Find GPG keys and owners:

Remove corresponding GPG Key.

If you need to replace old sysklogd with recent rsyslog on centos, `rpm -e --nodeps sysklogd` is rather kludgy as yum will remove initscripts, upon which most of the system is dependent on. However, it is possible to install and remove via the yum shell in one go, which resolves the dependency issues.

/proc/cpuinfo has the info you need to identify the number of processors, cores and threads.

To get the total number of processors/cpu cores:

Total number of physical cpus:

Number of cores per cpu:

To check if hyperthreading is enabled:

If siblings is a multiple of cores then hyperthreading is enabled.

Recently, I have come across the mentioned bug during a password reset of Trac:

Reference bug trac-hacks.org/ticket/3233

This looks like an issue when set to "Force users to change passwords after a password reset?". I changed the config to not force the password change.

Also removed the session attribute in reference from the trac database:

If you probe on varnish service status to check if it is alive, this can be setup as below. Put it in "vcl_recv" block:

Then check for http "200 OK" in the response code.

The "H" option in both ps and top lists the threads:

Examples:

For port redirection to work inside OpenVZ containers, ipt_REDIRECT kernel module needs to be loaded in the host. Edit "/etc/sysconfig/vz" and add it to the IPTABLES list.

This should then allow to redirect ports. So if you need to proxy existing apache via nginx or lighttpd and you do not want to switch apaches' default port 80, then the below rules will do the appropriate redirection to port 81 where nginx/lighttpd server is listening, serving static content and proxying to apache for dynamic content:

Where 192.168.10.2 is the internal IP resolver of domain/host.

If you are switching out static content that have gotten cached in nginx, the head of the cached files usually stores the file path that can be greped for and the file removed. One you hit the url again, it will recreate the new cached file at the same location.