Jump to content
Compatible Support Forums
Sign in to follow this  
news

ostree 2012.12

Recommended Posts

http://git.gnome.org/browse/ostree/commit/?id=2769ff36ffea37ab100008c028acf3d55a621bd3

http://ftp.acc.umu.se/pub/GNOME/sources/ostree/2012.12/ostree-2012.12.tar.gz

 

There are two major changes in this release. First, there's the

introduction of a new "archive-z2" type. This supplants the previous

"archive-z" (which ostree no longer reads). This repository type is now

used on http://ostree.gnome.org.

 

The major advantage of this is that content objects are compressed (just

with gzip), which translates into pretty significant download size

savings as you might expect. However, we still want to do the magic

hardlink trick for checkouts to keep buildroot construction cheap, so

there's a cache of uncompressed objects.

 

What's even more exciting in this release is that ostree-pull is fully

asynchronous and multithreaded. This is *huge* win for download times.

Especially on a warm disk cache, you can pull small deltas (e.g. just an

updated gnome-shell build) in *seconds*. Yes, seconds. Deployment is

unfortunately still mostly bottlenecked on gtk-update-icon-cache =/ I

also need to make the checkout of /etc asynchronous/threaded.

 

In the big picture, do keep expecting commands to be renamed, options to

change, etc. OSTree is still young. You have jhbuild and

rpm/dpkg/portage/whatever to fall back on, so I want to take at least

another year to battle-test the system and ensure things like the

on-disk filesystem format are *right*.

 

Colin Walters (32):

admin: Prepare /var/log/dracut.log when creating initramfs

admin: Install grub2 config file if we detect /etc/grub.d

pull: Asynchronous metadata fetch

pull: Merge metadata and content fetch phases

pull: Ensure queued filemeta requests don't starve everything else

pull: Stage metadata objects asynchronously

pull: Stage content asynchronously

pull: Ensure worker queue finishes if we unhold an empty queue too

core: Only do devino scan on commit (speeds up pull)

pull: Print elapsed time

core: Add dynamic uncompressed object cache for archive-z

repo: Avoid race condition in threaded checkout with symbolic links

core: Ensure file data is synced to disk when checking out via non-hardlinks

core: Make mkdir -p function safer

core: Fix regression in previous commit

core: Threadsafe recursive direction creation is harder than it appears

core: Prefix some error messages

core: One more minor tweak to recursive directory creation

core: A few more error-prefixings

core: Fix case where we hardlinked successfully but thought we didn't

checkout: Drop uncompressed object GC for now

core: Call fdatasync() before we rename()

core: Rework archive-z mode to have header be uncompressed

core: Fix devino speedup for archive-z

core: And one more race in multithreaded mkdir -p

core: Use linkat() rather than link() in threaded checkouts

core: Ensure correct file data is synced in non-overwrite case

core: Check out files before directories

core: quiet a compiler warning

libgsystem: Update

admin: Add new pull-deploy command

Release 2012.12

 

Jasper St. Pierre (1):

ot-main: Ignore lt- prefixes in binary names

 

Matthias Clasen (1):

Make ostree admin --help output more useful

 

 

 

_______________________________________________

 

Share this post


Link to post

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
Sign in to follow this  

×