Author: Paul Tomblin

Another glitch…

After my wipe and reinstall experiment, there are still a few little glitches (above and beyond the fact that I’m still running my RAID1 in degraded state because the disk it’s supposed to be mirroring to is being replaced). One of the most annoying was my hourly “rsync to the external hard drive” – my own home grown equivalent to “Time Machine”. About once a day it was freezing up, usually during the first big access to the drive, which is a “rm -rf” on the destination directory. And once it froze up, it really froze up – there would be dozens and sometimes hundreds of processes that wouldn’t end and couldn’t be killed, most the hdparm and smartctl commands that munin issues to detect the drive status and temperature. The only way to get those processes to end was the turn off the external USB cradle. This didn’t happen before, so I don’t know what’s going on. But I suspected that something is being more aggressive about spinning down idle external drives, so I issued the command “hdparm -S 0 /dev/sdd” and that seems to have solved the problem.

Other glitches – the dvd player application exits as soon as you start it. Also, I just had a notification in the system notifications area that there were upgrades to install, so I clicked it and it brought up the updater app, but when I clicked “Install all updates” it told me it couldn’t because I hadn’t authenticated. I’m pretty sure it’s supposed to prompt for your password and authenticate you first. So I installed them from the command line using aptitude. Nothing I can’t live with.

Well, that didn’t last long

A few hours after my last post, as I slept not very soundly, the hard disk that kept dropping out of the RAID as it rebuilt dropped again. This morning, I booted in the the Seagate “SEATools” disk and gave it a full on test, and it found a metric buttload of bad sectors. Not good on a brand new disk, so off to the RMA process I go. The machine is up and running again, with /dev/md128 running as a “degraded RAID 1” (ie a single drive with no mirror).

Does anybody else find it pretty shitty that the vendor can ship you a defective drive, and then make you pay for the shipping to return it? Or that they won’t ship the replacement until they get the old one back? Seems to me that they could ship a replacement immediately with the proviso that they’ll charge your credit card for the drive if you don’t return the old one.

A new start

I decided I needed some more room on my home Linux box, so I bought a couple of 2TB disks. I also decided it was time to do a wipe and re-install because the box has been upgraded many times and there are several things that just don’t work right any more. So I burned a bunch of different distros onto CDs and experimented with them, and decided that Kubuntu is the best combination of beauty and power. None of the other distros except Debian allows you to set up lvm on RAID-1 while installing, and Debian looks like a bit of a dogs breakfast compared to Kubuntu, even in KDE.

The two new disks are so big that as it is, I’m only using them for the new installation, so the I can mount the old system under /old_system and I can compare everything to make sure I haven’t lost any important configuration. But eventually I’ll be able to delete those partitions and add them back to the free pool in the LV and use them to expand any of the actual in-use partitions. I love lvm.

Installation has gone ok, I’ve got about 3/4s of the things I need to get working again working again, but I’ve hit the usual string of snags. The biggest is that one of the two new disks keeps dropping out of the RAID1, so I have to keep adding it back in. I’ve rejiggered all the cables as best as I can, and hopefully it will rebuild overnight and it will be ok. Fingers crossed, I’m off to bed.

Is this even possible?

Ok, here is the current situation: My machine has two identical 500Gb drives, and two identical 1Tb drives. When I first set it up, with just the 500Gb drives, I was using a version of Linux that wouldn’t boot from mirrored lvm partitions, so I set it up

  • /dev/sda1 2Gb /boot
  • /dev/sdb1 2Gb /
  • /dev/sda2 2Gb swap
  • /dev/sdb2 2Gb swap
  • /dev/sda3 and /dev/sdb3 – turned to a mirror, /dev/md0 and then created a
    pv and a vg.

That was working fine for a number of years, and then I added the two 1Tb drives, which I created as a mirror /dev/md1 and added it to the vg.

At this time, I have more more than 500Gb free space in the vg, so I definitely could pvmove and vgreduce and get rid of that. I’d also like to somehow fix it so that / and /boot and swap are on lv and just boot from the two 1Tb drives. The reason I’d like to do that is that

  • a) It’s been bugging me for years that I’m still using physical partitions
    for / and /boot, so I can’t resize them and
  • b) I have two 2Tb disks sitting on my desk that I’d like to add, and I
    don’t think I have enough SATA ports for 6 disks and
  • c) these two 500Gb drives are ancient – they just hit their 5th birthday a
    few weeks ago.

So would it be possible to migrate everything off of /dev/sda and /dev/sdb without major surgery? Or am I going to have to bite the bullet and re-install everything?

Probably time to throw in the towel on the waypoint generators

For years now I’ve been providing aeronautical data for various programs – The Wayback Machine shows it existing in 2001, but I’m pretty sure I was running it before then. The site hasn’t changed much visually since then, since I concentrated on providing good data rather than prettying up the site. But I thought I was providing a good free service and it was worth it to people. I didn’t really push it hard, but I did politely ask for donations, through Paypal and at one time through the “Amazon Honor System” until Amazon killed that. And for a long time, I made about 1/4 to 1/2 as much through that as it cost me in hosting and data costs. But that hasn’t been true for a long time. I just looked at my Paypal history for the last 12 months, and I got two donations of $25 each, one of $3.32 and one of $1.13 (yeah, go figure). $54.45 in a year. That’s it. That doesn’t keep me in thumb drives.

Add on to that the fact that my source of world-wide data, the DAFIF file, hasn’t been updated since 2006 and it’s increasingly been a “US data plus some airport data from here and there”, and I’m worried about presenting this to the world as useful data for anything. I’ve always staunchly maintained that this is just to save you from a bit of data entry for flight planning and not a substitute for official data sources, but it’s just embarrassing to think somebody is going to be looking for navaids and waypoints in Europe, say, and not finding anything that’s still in use. At least airports don’t tend to move around much.

And because the data I have is getting stale, I keep hoping that people would provide feedback on the data I have, offering to provide data for their area or just letting me know if something is wrong. At one time I had a couple of people sending me their data sets which I incorporated into the database. (Although I did have one set of data for Australia where all the airports had the sign of their longitude reversed – I’m pretty sure I purged the last of that.) That was actually more important than the donations to me. It showed people were using the data, and cared when it was wrong.

So I guess what I’m saying is that I’m reconsidering if it’s worth doing this. I’ll still maintain the data for the CoPilot iOS app, that’s a separate database and thanks to Laurie Davis I get world wide data from Eurocontrol that that. But next time I get a data dump from the FAA (September 19th), I’m going to give some thought over whether to load it into both database, or just the CoPilot iOS one.