Marc's Public Blog

I think the only minus we had was being tired at the end of the day. If that's the worst that happened, it was a great day :)

drivers meeting

And here are the cars I drove in sequence:

After 3 cars, it as time for lunch and a bit of rest :)

And I was up for the Mercedes SLS. I expected it to be a heavy pig with a lot of grunt, but I was pleasantly surprised. It actually felt as nimble as the F430, and had more power. I actually liked it quite a bit:

While last year I was a bit disppointed by the shorter loop compared to my first time, this drive was awesome. We were all beat by the end of the day, and we got a real chance to try each car on an equal footing. It was well worth the time and money. Thanks Ruch for setting up this special drive for us.

Soon before he left, we when didn't know how much he'd live still, Jennifer wanted to give him a little tour of our yard with a made up leach. He wanted to go there, but due to racoons, poison oak he could bring home, and more, that just wasn't practical for him or us, so she gave him a supervised walk for him to enjoy the plants and grass:

On his last day with us, I took a couple of pictures. By then he had lost a fair amount of weight already:

After that, we had to do a full and thorough house cleaning:

We'll miss him, he really was a good cat without an attitude and wanting nothing more than companionship, being pet, and sitting on your lap or next to you.

Plenty of flower art:

And other things:

How to manage a btrfs filesystem made out of multiple dmcrypt'ed drives

If you are using raid0, raid1, raid10, raid5, or raid6 with btrfs and you want your filesystem to be encrypted, you need to encrypt each device seperately but later you'll want a script to decrypt all those devices.
This can be done with /etc/crypttab, but I don't personally use it for arrays that I turn off to save power. You can use keyscript= in there to feed a script that will provide the decryption key, but I wrote my own script to tun the disks on, locate them by disk ID, decrypt them, and mount the resulting partition.

If you are planning on using Raid5 or Raid6, you'll also want to read this page.

For the mount to work, you of course have to create the crypted device and filesystem first. Here is a recommended way:

cryptsetup luksFormat -s 256 -c aes-xts-plain64  /dev/sda4
cryptsetup luksFormat -s 256 -c aes-xts-plain64  /dev/sdb4
cryptsetup luksFormat -s 256 -c aes-xts-plain64  /dev/sdc4

cryptsetup luksOpen /dev/sda4 sda4_crypt
cryptsetup luksOpen /dev/sdab sdb4_crypt
cryptsetup luksOpen /dev/sdac sdc4_crypt

mkfs.btrfs -d raid0 -m raid0 -L btrfs_pool /dev/mapper/sd[abc]4_crypt

After reboot, the idea is to avoid the luksOpen steps and adapt to whatever device names those drives could come up under, and this is what the script below does.

Here is the script, start-btrfs-dmcrypt, for which I'll paste a most likely outdated copy here:

#!/bin/bash

# Example script to decrypt a bunch of drives and then mount them as
# part of a btrfs volume.
# 
# By Marc MERLIN <marc_soft@merlins.org> / 2014/04/29
# License: Apache-2.0

# Get these from /dev/disk/by-id
DRIVES="
scsi-SATA_Hitachi_HDS7230_MN5220F323S79K-part1
scsi-SATA_Hitachi_HDS7230_MN5220F325UZMK-part1
scsi-SATA_ST2000DL003-9VT_5YD6MH88-part1
scsi-SATA_ST2000DL003-9VT_5YD70NHX-part1
scsi-SATA_WDC_WD20EARS-00_WD-WMAZA0374092-part1
"

# The label name of your btrfs filesystem (mkfs.btrfs -L btrfs_pool)
LABEL=btrfs_pool

NUMDRIVES=$(echo $DRIVES | wc -w)

die () {
    echo "$1"
    exit 1
}

pwd="$(yourscript that returns crypt key)"
if [ -z "$pwd" ]]; then
    echo -n "Decryption key? "
    stty -echo 2>/dev/null
    read pwd
    stty echo 2>/dev/null
fi
[ -z "$pwd" ]] && die "Didn't get a decryption key"

# Here you can run a command to turn the disks on if they are on an
# external power outlet.
# turn-disks-on-cmd
cd /dev/disk/by-id
for i in 1 2 3 4 5 6 7 8 9
do
    if [ $(ls $DRIVES 2>/dev/null | wc -l) = $NUMDRIVES ]]; then
	break
    fi
    sleep 10
done
# This is useful if the disks were just turned on.
/etc/init.d/smartmontools restart

for i in $DRIVES
do 
    dev=$(ls -l $i | awk '{print $11}' | sed "s#../..#/dev#")
    [ -z "$dev" ]]  && die "Couldn't find device for $i"
    echo "$pwd" | cryptsetup luksOpen "$dev" "crypt_$(basename $dev)" || die "Couldn't decrypt $dev"
    echo "decrypt $dev"
done
btrfs device scan
mkdir -p /mnt/btrfs_pool
mount -v -t btrfs -o compress=zlib,noatime LABEL=$LABEL /mnt/btrfs_pool || die "Couldn't find btrfs $LABEL"

The storm didn't quite follow the forecast, it feel harder earlier and then didn't deliver friday night like it was supposed to, but by then so much had fallen that it didn't matter.
In hindsight, Johannes and I should have arrived for Friday, noon to get some of the fresh cleaning tracks action, but eh, getting there for saturday morning, wasn't that bad.

More people than I expected showed up, but given the forecast, it wasn't surprising:

Unfortunately the weather got a bit warmer than it should and much of the powder was damaged by the end of the day (some even re-froze), but still, it was deep powder and good snow for the hours it lasted

the kiddy side had been closed, so we had to ride a shuttle back to the front side

pretty narrow, and many rocks...

Due to ice, I slid down on top of these rocks, and jumped feet down and on my butt over the rock to land onthe other side. I got lucky, could have been worse...

A big thanks to Johannes for getting us there and back and Jim for having us at his condo.

How to cancel a btrfs scrub that won't cancel

This is fixed as shown below:

Problem:

gargamel:~# btrfs scrub start -d /dev/mapper/dshelf1
ERROR: scrub is already running.
To cancel use 'btrfs scrub cancel /dev/mapper/dshelf1'.
gargamel:~# btrfs scrub status  /dev/mapper/dshelf1
scrub status for 6358304a-2234-4243-b02d-4944c9af47d7
        scrub started at Tue Apr  8 08:36:18 2014, running for 46347 seconds
        total bytes scrubbed: 5.70TiB with 0 errors
gargamel:~# btrfs scrub cancel  /dev/mapper/dshelf1
ERROR: scrub cancel failed on /dev/mapper/dshelf1: not running

Fix:

gargamel:~# perl -pi -e 's/finished:0/finished:1/' /var/lib/btrfs/*

Verification:

gargamel:~# btrfs scrub status  /dev/mapper/dshelf1
scrub status for 6358304a-2234-4243-b02d-4944c9af47d7
        scrub started at Tue Apr  8 08:36:18 2014 and finished after 46347 seconds
        total bytes scrubbed: 5.70TiB with 0 errors
gargamel:~# btrfs scrub start -d /dev/mapper/dshelf1
scrub started on /dev/mapper/dshelf1, fsid 6358304a-2234-4243-b02d-4944c9af47d7 (pid=24196)

It was a 2 day school, which started with getting us used to the cars, including shifting up and down without the clutch for me. My feet were too wide and I just could not put one foot on the clutch and another one on the brakes, they didn't fit. As a result, I got to learn to downshift without the clutch by "simply" braking and revmatching with the same foot.
We had some class time:

Nico gave us an intro to the car

Et voila, c'est facile! :)

Each car had full datalogging that the MX crew could download for us

The cars had a 4 cyclinder turbo engine that could be tuned up to 300hp. Lots of power for a car that weighs nothing. It can corner at 3G, more than your body tells you should be able to make a turn :)

We did some paddocks exercises to learn oversteer and understeer control as well as downshifting under braking:

We were a small group, just 5 of us:

On the 2nd day, we got to drive the track untethered:

After our last session, we got to see the data:

What I learned was I definitely needed to hit the brakes harder, which was challenging since I also had to revmatch at the same time for downshifting. Also, I wasn't full throttle in some corners where the tires could stick but my brain wasn't used to the idea of 2G cornering yet. In the end, I was 0.25s slower than the fastest person, doh...

To study for next time, here's a walkthrough of most corners:

Good time was had by all, this was a great learning experience in "different" cars. Looking forward to Level 2 :)