Uno-Code

This is a great tip a friend gave me the other day. The elog summary is great to get some of the warning messages and instructions, but you can have those email to you, and get more information! This is extremely simple.. and can really save some time debugging a emerge. Simply add this to your /etc/make.conf file

Obviously change the portage@yourdomain.com to your logging email account, and you'll get nice portage messages.

You ever want to see what your server is doing on boot up, but it's a remote server? You can log the boot messages to a file. I'm a little late to come across this, but it's still a mighty cool tip. First you need to emerge showconsole.

Next, we need to edit /etc/conf.d/rc and change this value to 'yes'.

On your next boot up, it will write to /var/log/boot.msg. Basically everything seen on console (except boot splash) will be written to this file. Very handy if you're having some boot issues.

The last two days have been maddening for me. On two servers, out of the blue, MySQL is freakin' and freakin' hard. It all started with corrupted tables, to memory preformance issues. I then jacked up some mem values, and now both servers are crashing. I think I might be dealing with two problems (second problem related to the first problem). I posted on the Gentoo forum, but another classic 'talking-to-myself' thread. Man, I miss the days when you'd get several reponses within minutes. My problems/questions seem pretty difficult lately, and that's probably a big reason for the lack of replies.

Below is my basic thread of the problem.. if anyone has any suggestions, please comment!

I'm getting desperate for some MySQL help on one of my servers. It started mysteriously lastnight. One of my applications (code has not changed in serveral months to a year), complained about corrupted tables. Viewing the logs, I was definitely seeing MyISAM corruption, which I could replicate by issuing a OPTIMIZE after the sql performed a UPDATE. Again, this code has been like this for many months. I commented out the OPTIMIZE and repaired the tables, and that seemed to have fixed the problem.

Hello All

I now have RSS feed subscriptions available. Get site updates via your RSS reader of choice to see what pain in the ass problem I'm currently dealing with.

Just click the the 'readers' widget to the right, or click this link:
http://feeds.feedburner.com/uno-code

So if you've haven't been following what I've been working on.. I've been working on a OpenVPN system on a Gentoo Linux firewall where remote users can connect and access files, but have the ability to authenticate against PAM as well as shared keys (dual layer authentication).

Most of the how-to documentation out there talk about a few iptable rules to put in place on your tap interface (link)....

Currently, my original 'how-to' on getting OpenVPN running with OpenVPN-GUI used the standard key based authentication. You can configure your client to password protect your connection, but I wasn't comfortable with that scheme. If a laptop is stolen, it would be possible (and not hard to bypass this password). Now, we can add a additional layer with pam and server side interaction.

We still need keys to create the tunnel and send our authentication, so the 'base' security layer is untouched, but now we're going to add pam to authenticate our user. This can be handy if you're creating multiple VPN users, simply removing their system account, will remove their ability to VPN, etc.

According to this thread I started, the 'unusably short session_id provided' issue I discussed earlier is related to a bug in openssl-0.9.8f. I've added 0.9.8g to package.keywords and I'm currently doing some testing.

For those that don't know how to do this, simply add this to your /etc/portage/package.keywords:

So the Pam upgrade wasn't too bad. I updated one box, restarted services, and rebooted to ensure that everything is working correctly. All you really need to do is update some of the files in pam.d.

When I did this, I also saw rexec, rlogin and rsh. I basically looked for service=system-auth and replaced it with the new format:

OLD:

NEW:

Hmmm, greated this morning with a pam upgrade. The einfo message looked somewhat hairy:

So with the latest update for openssl (dev-libs/openssl-0.9.8f), I've been seeing the following in my error_log with SSL sites: