[OmniOS-discuss] Bloody update for February 23rd (LONG!)

Dan McDonald danmcd at omniti.com
Thu Feb 23 19:06:30 UTC 2017


Hello folks!

Pardon the delay, I've been wrestling with what to do with install media, and I've decided to throw out Caiman and implement a simpler (and likely more primitive) installer using Kayak for ISO & USB.  It is under active development right now.  I HOPE to have it out in two weeks in time for the next bloody update. If I'm behind, expect the bloody update to also lag, as I really would like the next update to include Kayak for ISO/USB.  Also, the feedback from BSD Loader deployment has been either positive or none-at-all, so I assume everyone's okay with it.

Although the feature has been in bloody since the jump to Python2.7, we're officially committing to a new set of linked-image semantics for lipkg zones.  ipkg zones remain unchanged.  Per the prior thread: "On pkg(1) behavior in r151022", prior to the Python2.7 upgrade:

pkg update <pkg>
- Updates <pkg> on global zone.  No others, REGARDLESS of linked-image or not.  (This is considered a bug.)

pkg update
- Update all pkgs on global zone AND on any linked-image zones.


As of now, we are officially endorsing and embracing the following new behavior:

pkg update <pkg>
- Same as before (but now not a bug due to "-r"...)

pkg update
- Only updates global zone, and linked-image "If needs sync", where "needs sync" is defined by a parent-image dependency.  Currently no OmniOS packages have this dependency, but before this bloody cycle closes, some will.  (e.g. system/library, SUNWcs)

pkg update -r <pkg>
- Updates <pkg> on global zone AND on linked-image zones.

pkg update -r
- Behaves like "pkg update" on 014-020, updates all pkgs on global and linked-image zones.


The following pkg(1) primitives can now take the -r flag, and the -z (certain zones) or -Z (exclude zones) also:

	install
	uninstall
	update
	change-facet
	change-variant

So if you're an lipkg user, start including "-r" in your pkg update or install commands if you want the pre-Python2.7-wad behavior.  Many users have said this is an overall improvement, so it may, during the post-022/LTS releases, become the default "ipkg" behavior as well.  I believe Oracle intended these semantics for linked-images the whole time.

uname -v show omnios-master-05eff6d

illumos-omnios is master branch, commit 05eff6de1f5d0f947aef584ac1a3d2807198b841
	(LX bits last merged with illumos-joyent commit 384c07167202f048489a38c1fd2a2f8ac5a082bd)

omnios-build is master branch, commit 3d2982b721130970aa29068079386dd6fbc06378

Also new in this update:

- gcc44 is up to "il-4" to build the latest illumos-gate and illumos-omnios. This came out on the repo early.

- curl is up to 7.53.0, also an early arrival.

- OpenSSH is now up to 7.4p1, also an early arrival.

- bash is now up to 4.4p12

- nghttp2 is up to 1.19

- BIND is up to 9.10.4-P6

- SimpleJSON is up to 3.10.0 (and passes pkg5 tests!)

- Some ASCII art improvements to the OmniOS Loader screen (thanks Dominik Hassler)

- TCP keepalives now accepted with missing timestamps (helps with Windows interoperability)

- ZFS fixes
  - mmap() now updates timestamps.
  - 

- SMBIOS 3.1 support

- mpt_sas bugfixes

- Various LX fixes

- The illumos vmem() allocator now has man pages.

- Loader BIOS disk fix for disks > 2TB.

- IPv6 Atomic Fragments no longer exist (Per RFC 8021) unless you're using deprecated CGTP/Multirouting.

- Intel 40Gig Ethernet (i40e driver) now supports multiple transmit AND receive rings.


Happy updating! (And wish me luck on Kayak-for-ISO/USB...)
Dan



More information about the OmniOS-discuss mailing list