So last night in an abrupt turn of events, Watson the server would randomly shut down after a few minutes. After downloading a sensor program (for all Linux lacks documentation that understands there's a wide open space between "idiot's guide for beginners" and "advanced kernel things involving five thousand commands", God do I love that insta!downloading of useful programs) my cpu said it was running 128 C.
That, I thought, was a bit high into impossible.
Third time it happened, I went into the bios and watched, a little awed, as my CPU temperature rose and rose and rose and then shut down between seventy-four and eighty. I couldn't figure out how--there were two fans and I'd left the case open for ventilation and adding things. And until this time, the CPU had never done this.
So I googled for three long hours and could only discover that the linux program said celsuis, but read in fahrenheit, which whatever. I heard a lot about thermal paste. And still no explanation of why a server that had been fine hours before was now critically overheating. I finally ran it, waited for it to shut down, unplugged, and pried off the heatsink to finger check. One, ouch. Two, yeah, that was rather hot.
So I went to bed to cry. And then got up because a.) there is no crying in servers and b.) I was starting to dream about the server killing me. Googled again and randomly ran across a forum post that was about how to put together a computer from scratch with like, good instructions on the order in which to assemble one's computer; buried in it there was a line about the heatsink when not perfectly installed caused overheating.
(Point: you know, that would be very useful information in like, Intel's picture-only no-text what the hell were they thinking instructions. Maybe add a picture of what a perfectly seated heatsink looks like, too? Maybe?)
I was pretty sure at this point that nothing I did would ever ever work and I would die alone clutching computer parts and mumbling about thermal paste being of the devil. But. Sitting Watson in my lap, I unhooked every power and sata cable and wrestled my board out to slowly and carefully take the heatsink out, stare at my CPU in deep and abiding betrayal, put the heatsink back on and reattach it until I could hold the entire board up with it (I didn't shake it, but God did I want to. At that point, I was well aware I had acquired computer-related anger management issues). Then I double checked it. Three times. Reassembled, plugged in, went straight to bios and--right. Apparently, the heat sink came loose, because my temperature leveled out at 54 and stayed there. To point out, I still can't figure out how a.) it came loose and b.) how to check that when seriously, it was visibly fine. And I still stare at it because it looks pretty much like it did before. No idea how or why. Now it's on there and I want to superglue it into place except that would be bad. Or so forums seem to imply heavily.
Ubuntu Hates Me
Here is the thing I've suspected about linux for a while; it is not nearly as difficult as it looks. And the minute, the second you say that, the goddamn thing falls apart.
Strangely enough, I do like working from the command line a lot, and I didn't expect that. What drives me kind of crazy is the above on documentation--Linux and Ubuntu do great super beginner, but they also have a really bad habit of assuming that once you get the basic commands down, everything else will follow instantly, including the logic behind command structure and immediately jump to a darker shade of kill computer with fire because it really shouldn't take me combining the accumulated info from three ubuntu forums to figure out how to add a second drive, how to mount it, where to put it, and what that line in fstab actually means altogether. Sharing it, that I knew how to do. It also suffers from a bad habit of giving you three lines of commands and saying, use this! and then not explaining what each part actually does. I mean, I get they may be thinking they don't want to be condescending or anything, but seriously, be condescending if that means I don't have to write out my commands first on paper to make sure I get the order right after fishing the different pieces from different places.
After the heatsink issue was resolved, I opened ubuntu to find it wouldn't mount my second drive. At that point, I just didn't care--no overheating! Until I went back and remote accessed through my laptop, logged in, and stared in horror as it stated the home directory was empty and wtf?
(Note: /home is the directory had all the users and their directories, seven in total. WTF indeed.)
I did sudo find. I searched manually by cd'ing and staring bitterly. I tried to view the contents of the /home partition. I shook my fist (metaphorically). I cd'ed into home every five minutes like everything would magically reappear and considering events, who knew that was actually practical.
The /home vanishing directories self-corrected an hour later for no apparent reason after my blood pressure reached a healthy normal (I have low blood pressure, so my high blood pressure is just above that of a coma patient; last night I achieved a personal best of normally functioning human being) when abruptly, while trying to find a hidden trash can (yeah. Hidden trash can. Don't ask) it all reappeared in /home and just sat there. Mocking me.
IDEK. Was that like, a test or something? Of my sanity?
I'm actually not kidding about how documentation seems to work despite numerous guides that say beginner and lie. The first problem I ran into was an assumed knowledge of DOS, which for some reason it does not like to acknowledge is fairly depreciated outside of computer programmer circles. It also assumes at minimum a background in basic computer programming. I have the basic computer programming, but I never had to work in DOS outside of extreme panic situations regarding hard drive failure, so that was annoying. This would not be a problem if ubuntu server had the equivalent of a system restore function, so you could do a mass undo when you screw up; there are ways to do similiar things, but not as complete.
I got around that by creating three separate files for every configuration file I edited (four in some cases): x.conf.original, x.conf.current (copy of current working conf file, created and updated directly before making any changes at all) and x.conf (working configuration file). Optional fourth is x.conf.randomnamehere when I'm testing a lot of different changes and how they interact with each other and I want the equivalent of a system restore to return to how I had the system before I started, and every change is documented in the file and commented out (and with attached link to forum/site/place that contributed to the line of code). Also created a private log file where I list off any changes I've made to the system (hopefully in order of doing).
Useful things I wish had been documented somewhere:
Create all directories from the root if you plan to share it. I don't know why. I no longer care.
Do all file editing from the root. Same as above.
(One day I will find out the whys of the above. When I am comfortable enough with system performance to care.)
You will probably never know why a directory goes read-only for no reason despite permissions. Just go to root, create a new directory, move everything there, and delete the old one. Trust me when I say, no matter how tedious that is, it's just faster that way. Think of it as a really excellent way of practicing your ability to move and copy things. Lots and lots of practicing. When you're done, you will be able to do it in your sleep. And you may or may not have dreams involving it.
Print this and keep copy taped to your desk, computer, body, or somewhere you can stare at it regularly.
Bookmark this and have it ready at any moment for reference.
Having a fit of amnesia for how Windows works with files and directories will help a lot. Like, hugely.
If all else fails and it's been accumulated six hours working on something that isn't fixing but used to work, get your data moved to your local computer, format the disk, and reinstall from scratch. No, really. Again, it's just faster that way.
If the command they tell you to use doesn't work, even though it's like, in the documentation of that version that it should work, google until you find alternates and try them methodically. That is how I was finally able to restart samba after conf changes, because the way in the actual documentation did not work at all. But an older one from a completely different system did. No clue.\
Don't be scared to break it. Just watch a movie while you do the reinstall.
Testing this next for the concept of system restore sometime in the next two weeks. *crosses fingers*
Posted at Dreamwidth: http://seperis.dreamwidth.org/51798.html. | You can reply here or there. | comments