| 2017-11-16 12:29:14 |
Balint Reczey |
bug |
|
|
added bug |
| 2017-11-16 13:46:12 |
Balint Reczey |
livecd-rootfs (Ubuntu): status |
New |
Triaged |
|
| 2017-11-16 13:46:22 |
Balint Reczey |
livecd-rootfs (Ubuntu): assignee |
|
Balint Reczey (rbalint) |
|
| 2017-11-20 20:21:54 |
Balint Reczey |
branch linked |
|
lp:~rbalint/livecd-rootfs/xz-6 |
|
| 2017-12-04 22:19:03 |
Launchpad Janitor |
livecd-rootfs (Ubuntu): status |
Triaged |
Fix Released |
|
| 2017-12-05 11:59:45 |
Balint Reczey |
nominated for series |
|
Ubuntu Xenial |
|
| 2017-12-05 11:59:45 |
Balint Reczey |
bug task added |
|
livecd-rootfs (Ubuntu Xenial) |
|
| 2017-12-05 12:00:06 |
Balint Reczey |
livecd-rootfs (Ubuntu Xenial): status |
New |
In Progress |
|
| 2017-12-05 12:00:11 |
Balint Reczey |
livecd-rootfs (Ubuntu Xenial): assignee |
|
Balint Reczey (rbalint) |
|
| 2017-12-05 12:05:08 |
Balint Reczey |
description |
Switching from pxz to xz -0 in LP: #1701132 made the downloadable raspi2 image bigger, ~360MB which is a notable size increase from the original ~250MB while decreasing time the needed for compressing it.
The bigger compressed image is currently downloadable for Artful:
http://cdimage.ubuntu.com/ubuntu/releases/17.10/release/
Zesty's image is 232MB:
http://cdimage.ubuntu.com/ubuntu/releases/17.04/release/
To find the proper trade-off I ran compression tests on Xenial (xz ignores -T4 there):
https://bugs.launchpad.net/ubuntu/+source/livecd-rootfs/+bug/1701132/comments/5 :
...
rbalint@xenial-test:~/Downloads$ ls -al ubuntu-16.04.3-preinstalled-server-armhf+raspi2.img.p.xz
-rw-rw-r-- 1 rbalint rbalint 254621780 Nov 16 00:03 ubuntu-16.04.3-preinstalled-server-armhf+raspi2.img.p.xz
...
-0 -T4 81.56 73.64 1.46 4820 29.27 25.82 0.09 2252 329984
-1 -T4 103.68 95.14 1.67 10800 24.41 22.86 0.12 3180 298060
-2 -T4 134.05 126.22 1.76 18480 22.06 21.65 0.13 4232 286276
-3 -T4 170.57 162.72 1.72 33812 21.62 20.94 0.12 6188 275036
-4 -T4 322.15 317.83 2.10 50284 21.55 21.05 0.07 6136 258684
-5 -T4 407.15 399.93 1.92 97656 21.85 20.50 0.12 10412 247320
-6 -T4 478.64 473.32 1.95 97660 21.92 20.34 0.09 10348 245772
-7 -T4 493.97 485.52 1.99 191692 20.66 20.01 0.08 18548 240192
-8 -T4 512.28 506.08 2.20 380096 22.03 19.58 0.11 34888 229740
-9 -T4 652.71 559.20 4.78 691076 20.57 19.15 0.08 67700 224496
...
The tests on Bionic are still running.
Generating the raspi2 image is not part of autopkgtests at the moment, but may be periodically tested internally often and I can't tell the number of downloads either thus I'm asking for help to pick the proper compression setting for xz.
Looking at the Xenial numbers I believe -3 would result reasonably fast builds with a tolerable increase in size compared to xz -6's results. |
[Impact]
* Switching from xz -0 to xz (-6) for compression raspi2 image.
[Test Case]
* To test the change install updated livecd-rootfs on armhf and create images setting the following environment variables: ARCH=armhf SUBARCH=raspi2 PROJECT=ubuntu-cpc.
Notice that the xz-compressed rapi2 image is still created with xz's default (-6) compression.
[Regression Potential]
* In case of this change causes regression it would manifest itself as failing build for raspi2, failing build for ubuntu-cpc images or using other compression for raspi2 images than xz -6.
[Original Bug Report Text]
Switching from pxz to xz -0 in LP: #1701132 made the downloadable raspi2 image bigger, ~360MB which is a notable size increase from the original ~250MB while decreasing time the needed for compressing it.
The bigger compressed image is currently downloadable for Artful:
http://cdimage.ubuntu.com/ubuntu/releases/17.10/release/
Zesty's image is 232MB:
http://cdimage.ubuntu.com/ubuntu/releases/17.04/release/
To find the proper trade-off I ran compression tests on Xenial (xz ignores -T4 there):
https://bugs.launchpad.net/ubuntu/+source/livecd-rootfs/+bug/1701132/comments/5 :
...
rbalint@xenial-test:~/Downloads$ ls -al ubuntu-16.04.3-preinstalled-server-armhf+raspi2.img.p.xz
-rw-rw-r-- 1 rbalint rbalint 254621780 Nov 16 00:03 ubuntu-16.04.3-preinstalled-server-armhf+raspi2.img.p.xz
...
-0 -T4 81.56 73.64 1.46 4820 29.27 25.82 0.09 2252 329984
-1 -T4 103.68 95.14 1.67 10800 24.41 22.86 0.12 3180 298060
-2 -T4 134.05 126.22 1.76 18480 22.06 21.65 0.13 4232 286276
-3 -T4 170.57 162.72 1.72 33812 21.62 20.94 0.12 6188 275036
-4 -T4 322.15 317.83 2.10 50284 21.55 21.05 0.07 6136 258684
-5 -T4 407.15 399.93 1.92 97656 21.85 20.50 0.12 10412 247320
-6 -T4 478.64 473.32 1.95 97660 21.92 20.34 0.09 10348 245772
-7 -T4 493.97 485.52 1.99 191692 20.66 20.01 0.08 18548 240192
-8 -T4 512.28 506.08 2.20 380096 22.03 19.58 0.11 34888 229740
-9 -T4 652.71 559.20 4.78 691076 20.57 19.15 0.08 67700 224496
...
The tests on Bionic are still running.
Generating the raspi2 image is not part of autopkgtests at the moment, but may be periodically tested internally often and I can't tell the number of downloads either thus I'm asking for help to pick the proper compression setting for xz.
Looking at the Xenial numbers I believe -3 would result reasonably fast builds with a tolerable increase in size compared to xz -6's results. |
|
| 2017-12-05 12:06:21 |
Balint Reczey |
branch linked |
|
lp:~rbalint/livecd-rootfs/deps-xenial |
|
| 2017-12-06 10:27:55 |
Balint Reczey |
livecd-rootfs (Ubuntu Xenial): status |
In Progress |
Fix Committed |
|
| 2017-12-06 10:29:03 |
Balint Reczey |
livecd-rootfs (Ubuntu Xenial): status |
Fix Committed |
Invalid |
|
| 2018-02-15 19:46:04 |
Francis Ginther |
tags |
|
id-59ff10e7eddaf8d676d43eb3 |
|
