Posted by Trejkaz
Wed, 14 Jun 2006 00:08:00 GMT
I completed most of the assembly of the new media centre after work last night, finishing everything but the CPU and RAM. Insomnia struck at about 4am this morning, and since there was no point going back to sleep even if I could, I decided to head to the office early to finish the job before the normal working day started. The basic hardware is all in place now.
The newer CPU heatsinks being employed by AMD are really nice. I can’t count how many times I’ve been pushing down on a screwdriver, scared that it will slip or break, and smash some expensive piece of electronics. Now, you just have a little rounded lever, which wedges it firmly into place and locks onto the socket.
Lessons learned:
- It’s really hard to discipline yourself to install drives upside-down, even when you know they have to be installed upside-down.
- IDE cables really blow when you’re trying to save space inside a case. A SATA DVD-ROM drive would have been really nice, but for some reason there is “no demand” for them (a dongle costing more than $50 doesn’t count, because that would triple the effective cost of the drive.)
- The provided front-panel audio connectors will never correspond to the available pins on the motherboard (now that I have four computers where this is the case, I can be fairly safe to make this kind of assumption.)
Now I’m just waiting for an opportunity to start a Linux install on the box. Maybe that can start at lunch. Or maybe I’ll just have to wait until the work day ends again.
Remaining hardware I need is mainly an IR receiver. I’ve noticed that many video encoder cards come with them, so perhaps I should just use that approach. I do need something compatible with LIRC though, and I’m not yet sure how much that need will restrict my choices.
Tags computer, hardware, media | no comments
Posted by Trejkaz
Tue, 13 Jun 2006 06:09:00 GMT
Most of the parts for my new media centre arrived at the office this morning. The sweetest thing about it is the look of the new case. The Silverstone LC11 is a really nice piece of work, even if it is slightly larger than my current media box.
As far as assembly, it looks like the DVD drive is the only thing stopping me finishing it in one run. I couldn’t just buy the first black-fronted drive I found the other day because I had to confirm whether the drive door would fit the slot in the case. Now that I’ve seen the way the front drive door works, I’m fairly confident that it will fit, so I suppose a trip back to the local parts shop is in order.
The worst thing though? Knowing that real work has to be done first. I can’t start any assembly until the work day is over. ;-D
Tags computer, hardware, media | no comments
Posted by Trejkaz
Mon, 31 Oct 2005 00:55:00 GMT
I bought myself a nice new, 300GB SATA hard drive on Friday, as I’d finally saved up the money for it. On Saturday I eventually hooked it up, and started moving across the data which was outgrowing my older 120GB drive.
The copy process was taking forever to do from the console so I thought I would do it through KDE just to see the rate it was copying at, and it turns out it was doing an average of around 6MB/s (only slightly faster than copying a file across the network.) I was quick to blame the Linux SATA support at first, but then I started doing some more tests…
- Copying files from the old drive to the new drive = 6MB/s
- Copying files from the new drive to itself = 35MB/s
- Copying files from the new drive to the old drive = 30MB/s
What the? And lo and behold:
- Copying files from the old drive to itself = 10MB/s.
So, the old drive was on the way out and it took this new drive to tell me. SMART gave no errors nor warnings… as far as it was concerned, the drive had a clean bill of health.
The old drive, thankfully, is still under warranty, so getting it replaced was not an issue, though it would mean sending the thing back to the manufacturers. The real issue was that this drive had my boot partition and practically all my data on it, and it all had to be shuffled around.
After a couple of hours of indecision, I eventually decided to sacrifice my old 20GB backup drive to become the new boot drive (200MB /boot, 4GB swap, 800MB /, 15GB unallocated for now.) My old /boot and / moved across to that.
Then the new 300GB got split 40GB unallocated (future plan is to move my Windows dual boot onto that), 260GB LVM.
Ah, LVM… it saved me so much time. For /usr, /var, /home and /data were all on LVM, and it only took me one command to move them all off the old drive, and another command to remove the old drive from the volume group. After that, everything was fairly happy and I rebooted with the backup drive as primary master and the new drive as secondary slave.
I did attempt to shred the data on the drive, though I was only able to do about 5 passes before the thing would lock up (more proof that the drive is about to die) and gave up after that, disconnecting it permanently.
The only immediately irritating thing left is that my machine locked up overnight even while the old drive was disconnected. I hope that this is either due to the newer kernel I’m now running, or the newer NVIDIA driver (NVIDIA are already a good scapegoat, because their driver crashes so goddamn often.) I will update the kernel to a newer one, and perhaps downgrade that NVIDIA driver back to what I was using before which was known to be stable (it’s not like I’m running a cutting-edge card anyway.)
Anyway, incidents like this make software feel stable as a rock. And they really make you thank Linux for LVM because I would have wasted a lot of time buggering around with partitions if I weren’t using it. :-)
Tags computer, hardware, trypticon
Posted by Trejkaz
Mon, 07 Feb 2005 03:46:00 GMT
This might possibly be my first on-topic post ever on this weblog since it relates directly to Trypticon infrastructure. Amazing.
Our little server box at home was having some serious issues last week. Logging in from work was almost impossible at times, and towards Friday it seemed that logging in from the local network was just as futile. Packet loss was starting to top 60% even when pinging from another machine hanginf off the same switch. Complete madness.
Anyway, since it didn’t happen on the previous install which was on the machine, I had two suspicions…
- Something about networking in kernel-2.6.8.1 was screwy, or
- Something about the network card itself was screwy.
Not wanting to gamble, I figured I would fix both issues at once. The particular network card we were using to connect it to the LAN was amazingly a 10Mb/s card (must have been the one we used to run between the server and the cable modem, back when it used to do our routing,) so I switched it over to use the 100Mb/s card instead. That should have the obvious benefits. :-)
Aside from that, I had it compile a new kernel overnight (literally… it took somewhere on the order of eight hours to complete) and rebooted in the morning.
Everything was more or less fine after that, with only one obvious problem remaining. That problem is this: you can’t expect to run Apache, Samba, Postfix, and at least three other services on a machine with only 64MB of RAM, and expect all services to be responsive at the same time. :-)
So I’ve come to the conclusion that we need to replace the server even though it’s working again for now. It’s been running for a very long time now, my best recollection is getting it some time around 1998, so it’s lasted extremely long under the circumstances. The machine used to run Windows and first became a Linux server when I wanted to run Windows and Linux without the hassle of dual booting… back when I was still sceptical of Linux as a desktop OS (of course, I’m thoroughly sold on the idea of Linux as a desktop OS these days.)
So the new server, well… it doesn’t have to be shit-hot. My current thoughts look something like this…
- Cheap motherboard with everything on-board, assuming it has on-board networking that doesn’t completely suck. Two network interfaces is considered a bonus, but I don’t believe such a thing exists on any motherboards which still count as “cheap.”
- 2.4GHz Celeron or similar low-end, less heat-prone CPU.
- 256 or 512MB RAM. 256MB would honestly be enough right now, but with 512MB I might even be able to run my Java-based web server from it, along with everything else.
- 120GB hard drive. I will most likely stick to the standard Western Digital on this. I may be inclined to switch to using RAID-1 in the future though, so perhaps it’s a better idea to stick to 80GB on the initial disk, getting dual 120GB disks in the future.
- Reasonably quiet case and power supply, at least as quiet as possible without breaking the budget.
Ideally, it would be a completely fanless setup to ultimately remove noise, but I get the feeling that this isn’t possible without going overboard. The budget for this system is around $400, which I’m finding hard to meet.
So if anyone wants to suggest a way I can meet this budget, go ahead. :-)
Tags computer, hardware, trypticon
Posted by Trejkaz
Tue, 01 Feb 2005 03:00:00 GMT
It seems I’ve found a temporary workaround for my system’s USB2 issues. If I disable EHCI in the BIOS, everything works fine both in Windows and in Linux. So I’m inclined to believe that the motherboard is faulty, and right outside the warranty period, too.
But at least I can actually access my USB2 devices in the meantime, using the slower data rate. :-/
Tags computer, hardware
Posted by Trejkaz
Mon, 31 Jan 2005 06:08:00 GMT
Right at the start of the Xmas holidays, my 19” monitor burned out. Having a monitor burn out right at the start of holidays – the time it would be used the most – was a real pain in the butt. But worse was that the nearest repair place which handles LG’s repairs (and thus the only place I could get it done at without too much hassle) were just closing on that day. So I had to wait until half way through January 2005 to even send the monitor in to get looked at.
In any case, I was lucky enough to be able to borrow a monitor from work, and at least LG pay for all the shipment of the monitor back and forth until it works properly.
But here’s the strange thing: the monitor got to the repair shop, and somehow they can’t reproduce the problem. This is particularly odd, as the problem consisted of the monitor not displaying anything when switched on, and was accompanied by a lovely burning smell which was pretty hard to miss. Yet apparently, at their end, there was no problem.
So obviously somehow, something got jostled back into place during transit, which seems a little odd, as a burnout is about the last thing you would expect to “fix itself” in transit.
So they’re going to leave the thing on a day or so to see if it happens again. I can only hope that it breaks and they manage to fix it. What bothers me the most right now is that it might not break, and they might just send it back as-is, only for it to break again during transit… resulting in the same painful process over and over.
I hate hardware.
Tags computer, hardware, monitor