Posted by: Xjlittle
centos, grub, gui, headless, inittab, lilo, Linux, ssh, ubuntu, ubuntuserver, ubuntuserver upgrade, zimbra
Yes you read the title right. Ubunutserver is unbelievably stupid.
Ok let’s take it from the top. I had Ubuntuserver 7.0.4 installed as my WebDAV secure server, my son’s baseball team’s website, samba and so on. I run this server headless and without a gui. Note here that the normal install is without a gui and is touted loudly by the folks who develop ubuntuserver. This is all well and good.
Lately it’s been on my mind to look into and install a collaboration suite. Having searched around I finally landed on Zimbra.
After reading the documentation I see that Zimbra is certified on Ubuntu 8.04 LTS. Ok, well this is as good of a time to upgrade as any. This is where the fun (read stupidity) begins.
I ssh into my server as normal. I do all of the pre-distribution upgrade stuff, reboot and begin the distribution upgrade. The first thing I get is a message saying something to the effect of “We do not recommend using SSH while you upgrade. If you lose a connection it is difficult to recover.” Ok. I’m not doing this over a WAN link but on my local LAN. I haven’t had any network outages since I don’t know when, the sun is shining so no power outages looming. Things are looking good.
The upgrade proceeds smoothly. On several occasions I am asked if I want to replace any of my configuration files. This include the sshd configuration file. On all of these I take the default which is “Do not replace, I want to keep the configuration file that I am currently using” . That’s paraphrased but that is what it meant.
The distribution upgrade finishes. I get a nice message saying everything went well and asking me if I want to reboot. Of course I do. I type y and off we go. After waiting the appropriate amount of time I try to SSH back in. I am pretty excited at this point about getting my Zimbra install started.
Uhoh. No such luck. The dreaded “ssh: connect to host 10.10.15.105 port 22: Connection refused” message. What?!? Are you kidding me? No way. I try pinging the server. Yep, network came back ok. I try again. Nope. And several times after that. Now I’m writing this rant.
You see, here is why I think this is stupid. The server installs without a gui. That implies that no one is going to use this for a workstation. If no one is going to use it for a workstation, why then, should it have a monitor on it? If it doesn’t have a monitor on it, why would you not want someone to upgrade using SSH? Would the Ubuntuserver people prefer telnet instead? If the upgrade process is smart enough to know that you are upgrading over SSH then why not start SSHD on the reboot? If that is in fact the problem since I haven’t dragged a monitor over there to see.
I made a decision not long ago not to scrap my Ubuntuserver in favor of CentOS. I am starting to regret that decision. You see there is at least one other annoyance with Ubuntusever that I don’t find appealing either. This is the fact that they have decided that using an inittab is old fashioned or something. While this doesn’t have much affect on a server, at least one not running a gui, have you tried to boot into runlevel 3 lately? It’s not nearly as easy as it is with a GRUB boot loader where you can edit it, type in 3, hit enter and b and boot to runlevel 3.
Ok, I’m done. Next free time that I have I’m putting CentOS on there (as soon as I drag a monitor to the server and get SSHD started that is).