Kristopher Ives's questions

Currently we keep a DNS rule for every client that has mail.domain.com and this helps us redirect their mail to another server like GMail if they don't want to use our server. The problem is that when these domains change IPs or their individual SSL certs expire they can be locked out of their e-mail or have to confirm security exceptions they don't remove later.

To summarize we have something like this as DNS records:

domain.com         1.2.3.4
mail.domain.com    MX 1.2.3.4 <-- sometimes changes for people with GMail

The plan is we move our clients to using a single hostname for connecting to the mail server. So instead all clients would always connect to mail.secureinternetthing.com - This seems to work fine for new clients, as we tell them the new address to use, but for older clients we want to avoid clients having to change their settings.

If we change the MX record mail.domain.com to go to the same host as mail.secureinternetthing.com will our clients reject this?

I have a CentOS server running WHM that uses FastCGI (mod_fcgid) running PHP 5.2.17 on Apache 2.0 with SuExec. When I start Apache it begins fine and serving requests. If I run ps on the terminal as root I see the php processes and they are owned by their httpd parent processes.

After X amount of time - different from time to time, not much longer than a few hours typically - the server will begin spawning PHP jobs owned by the init process ID (1)

Example of good listing:

12918 18254 /usr/bin/php
12918 18257 /usr/bin/php
12918 18293 /usr/bin/php
12918 18545 /usr/bin/php
12918 18546 /usr/bin/php
12918 19016 /usr/bin/php
12918 19948 /usr/bin/php

Then later something like:

1  6800 /usr/bin/php 
1  6801 /usr/bin/php 
1  7036 /usr/bin/php 
1  8788 /usr/bin/php 
1 10488 /usr/bin/php  
1 10571 /usr/bin/php  
1 10572 /usr/bin/php 

The php processes running owned by (1) never get cleaned up. Why would these processes be running? We don't use setsid or anything beyond basic PHP in the code this server is running.

Cheers & Thanks

We setup a new Rackspace to shift capacity from our an existing server we have that is overtaxed. During this process we setup a new SSL cert using the command line instead of WHM / cPanel tools that my boss typically uses to do that.

I followed this Ubuntu wiki article:

https://help.ubuntu.com/11.04/serverguide/C/certificates-and-security.html

My problem is that when I go to the domain:

https://stoysnetcdn.com

It doesn't resolve very quick at all the first time. It seems to always resolve quickly with HTTP, but the first time I access it on a machine it seems to be slow at doing something with verifying the certificate.

How can I diagnose why this is being slow so I can fix it?

Thank You!

Our server is performing decently, but when backups or other scan processes run it will tank the entire server. Something like clamd will run and scan many files. While we expect slow performance, it's killing our cache and the end result is we don't get a system that can do anything.

Is there a way to disable disc/disk caching for a single command? The idea would be to run it like this:

# ./nocache clamd

Then while running clamd it would not thrash the primed cache while reading all the files on the system.

I have needed to duplicate a FastCGI + suPHP, and I have re-created the setup but unlike the CentOS one my error_log files are all in /log/apache/ instead of the directories where the script triggered the error. How do I configure PHP to write all error_logs to ~/error_log?

We have about 200 databases and since upgrading from 4.1 to 5.1 we've noticed that PHPMyAdmin takes a while to load the left navigation sometimes. Is there anything we can do to make this quicker?

Thanks!

On a dedicated server we use the shell is setup differently between accounts, which is quite irritating. When I ssh into most users I get:

[user@machine some_dir]$

However, with some accounts I get:

bash-3.2$ 

I keep a lot of shells open to various machines, and there are a lot of different accounts. I have the ability to either make everyone have the same ~/.bashrc, but I also believe there is a global config somewhere in /etc

What's the fastest way to make sure I get the same shell whenever I ssh?

A client website was recently compromised, and we're analyzing the damage and exposure from the attack. We've found a script that is "calling home" to a URL we can easily find in a log, but is this logged, and where?

/usr/local/apache/logs/access_log

Are outbound connections (ones from file_get_contents with a URL wrapper, like HTTP) logged in this file, or anywhere?

I want to see similar output like:

ls -l /etc/php5/php.ini

Which would give me:

-rw-r--r-- 1 root root 40344 Jun 14 14:33 /etc/php5/php.ini

Except the file in question is a directory, and doing ls -l on it would list it's contents.

How can I have it give me that permissions/info for a directory, or is there a better way?

After some server changes our PHP configuration got nuked, which caused a client's large mailing list to time out partially through about 4,000 addresses or more. We want to avoid sending a duplicate mail as much as possible, and we have the original mailing list.

The mail was sent using PHP's mail() function, which uses sendmail, which says it's using Exim. I've seen the command:

sendmail -bp

Which gives me a list of mail that appears was sent. This list isn't very big, and could be accurate (meaning the script timed out after sending only a hundred or so e-mails)

Is sendmail -bp the best way to get this listing?

EDIT: I can't find any mail.log file in /var/log/mail.log or via locate mail.log, but WHM seems to have some information on this.