Hi guys,

I'm doing a short survey on sysadmin's job roles. I heard that sysadmins never upgrade their linux servers, and I'd like to know whether this is true. So if you are a sysadmin can you take some time clarifying my doubt? Thanks!

1. Which distro do you use for your linux server?
2. Do you periodically apply security patches?
3. Have you ever upgraded the linux kernel?
4. Have you ever upgraded the linux system itself? (something like upgrading a Red Hat 9)

Thanks!

lazycai

I usually wait a couple months for the big bugs to get fixed, and then I upgrade my Fedora install, a procedure which always begins with headaches and ends with my site running correctly again under the new Fedora version. Even yum update can cause problems. I make backups, so that's why I feel safe enough to do this. Most sysadmins I know run either older or old-ish Linux distros like Fedora 8. I found one guy who runs FreeBSD 4.x!

1 members found this post helpful.

The Linux distribution you use for your server also determines how you carry out your updates. Fedora is a great Linux distribution, am actually using Fedora 12(Constantine) on my laptop but I never advice its use on servers. Reason. You'll wake up tomorrow only to find out that the repository servers are no longer available giving you two options:
1) Do your own upgrades and software as required
2) Format and install the current version.

In servers, I'd advice that you use a free Linux OS that is stable such as FreeBSD, Debian or CentOS. They would appear old and boring but they surely are stable and this is what we need for our servers. Right?

1 members found this post helpful.

I use Red Hat and CentOS for my Servers and for me, updating my servers is not a problem because i am having a true and working test environment for all my servers and applications (even for an Oracle RAC Cluster), and first I test the updates and patches on the test servers, when enough testing is done, then i deploy those updates on the Production Servers (But backup and disaster recovery are in place).
For building up test environments, I have got enough computing resources and I have utilized the Virtualization Solutions as well.

1 members found this post helpful.

I do update my servers: Centos 5.x+FreeBSD 7/8,Centos just went a app have big bugs, FreeBSD every 2 weeks.

Like saifkhan123 first test in a vm environment latter to production, with good backups.

See u!!!

1 members found this post helpful.

FreeBSD 4.x, not more surprising than a Red Hat 8 or 9...
so it is quite common to run old-ish Linux distros?

I only update when there is a real need for the update. When I update or upgrade, then it has to go through DTAP (development, test, acceptation and production). And everything is done after amking full backups and servers processes are migrated to other nodes in the cluster.
I am using CentOS and Debian on my servers. These are stable distro's with a good support from the community and repositories that will be available for long time.

1 members found this post helpful.

nice!

Of course, that is silly.

RHEL4, RHEL5, and FreeBSD6 (not Linux, as you know).

Security patches are applied immediately on non-production systems, and soon (within change control procedures) on production systems.

Kernel updates are included in the previous question / answer...

For RHEL, we "upgrade" across minor versions, but for a move to a new major version it is a complete server migration. For FreeBSD, I upgrade across major / minor versions as needed. (I need to get to 8 one of these days.)

1 members found this post helpful.

Hello,

Whenever a patch or update comes out I make it a point of installing it ASAP in our testing environment because most of our applications are available through the internet and just like previous posters, after the necessary testing, deploy it to production environment.

Since 95% of our servers are virtual ones (VMWare) and the disks are stored on a NetApp NAS with snapshot (4 times a day), I don't have to worry much about recovery in case of failure.

I have Debian 5 (mixed 32 and 64 bit) installed on 16, Debian Squeeze on 2 and RHEL5 on 3 servers.

I have upgraded the kernel (and complete Linux upgrades from Etch to Lenny and Squeeze) on several servers since they were not updated and old(ish) versions when I entered the company where I work.

Kind regards,

Eric

1 members found this post helpful.

veeeeeerrrrry helpful, thanks!

1. Always try to use a tested stable distribution like CentOS or Debian. And never something like Fedora which is a testbed.
2. Yes. Security is utmost important.But only from the sources reliable.
3. Yes. I do if need arises.
4. Yes. Usually upgrade my CentOS servers. Recently did it from CentOS4 to 5 and now its running 5.5 (I hope its just an example. else who is going to use RH9 for servers?)

1 members found this post helpful.

1) @work: All our servers run CentOS. For my own stuff I use Debian or Ubuntu LTS (though I'd prefer Debian server there too...)
2) Security patches are applied asap, after testing in test environment.
3) Yes, often enough
4) CentOS: point releases are easily done. Major releases usually are done through migration. However with CentOS 4 still being supported for a while, some of our 4.8 boxes won't see a migration just yet (unless hardware requires it).
Debian: Easier to upgrade to a new major release, so for my own stuff I do that. Doesn't come without hassle though, and in a production environment I'd suggest to go via migration too.

1 members found this post helpful.

I see I missed a question: I compile my own custom kernels, based off a config file called ratfood. Much like the Fedora kernels, but it lets me use my beeper, along with other things. I take the cheap way and compile with make O=kernel instead of the fedora way.

1 members found this post helpful.

So we see many Debians & CentOSes...

to chaitanya:
One of my friends has an RH8 running which hasn't been restarted for 1500+ days...

to bendib:
thanks for sharing but I'm just curious why do you use a Fedora as your server os?