|
Replies:
1
-
Last Post:
Jul 22, 2006 3:04 PM
by: dclarke
|
|
|
Posts:
312
From:
Burlington, MA
Registered:
6/13/05
|
|
|
|
Announcement: BrandZ build 43 now available
Posted:
Jul 22, 2006 2:54 PM
|
|
Greetings,
It's been quite a while, but a new BrandZ build is now available. This
build brings us into sync with nv43, fixes more than the usual number of
bugs, and introduces our new installer.
The new installer will continue to support tarballs, but we can finally
install Linux zones directly from the CentOS 3.x install media - either CDs
or ISO images. The final release will also support Red Hat 3.x, but the
underlying mechanism is the same.
This is likely to be our final release prior to integrating with Solaris,
so we are getting pretty close to what we consider production-quality.
There are still a number of bugs in here, but it's no longer something you
need to be scared of.
Since we are getting ready to integrate into Solaris, we would really
appreciate any and all feedback you can offer. If there is some flaw that
has been bugging you but which you haven't mentioned yet, now would be a
perfect time. If you've been waiting for BrandZ to stabilize before taking
it out for a test drive, your wait is over.
For those of you who have been with us for a while, there is one user
interface change you should know about. The zonecfg syntax for creating a
Linux zone has changed from:
create -B lx
to
create -t SUNWlx
Other than that, everything should work exactly as you are used to - only
better.
Nils
Bugs fixed:
-----------
6300832 atd fails to start in linux zones
6314815 /native/usr mount exposes too much
6321840 getsockopt() should follow linux man pages to set the corresponding
errnos.
6327393 linux needs undocumented bits cleared in mode field of struct sem_perm
6327433 sendmsg() should check validity of arguments and set correct errno.
6327947 sendto() should check validity of arguments and set the same errno as
linux does.
6328590 setgroups() doesn't follow linux man pages.
6335787 NGROUP_MAX define differently between linuxzone and global zone
6341428 lx brand zone doesn't shutdown or reboot
6348989 mmap fail with MAP_ANONYMOUS
6362285 brandz msg operations can likely be moved to user space
6365610 lx_install sorts ISO image names incorrectly resulting in an install
failure
6365617 lx_install should do more cleanup after a failed installation
6365774 I have a dream that someday iso installs will actually work
6365774 I have a dream that someday iso installs will actually work
6366049 BrandZ installer should be able to launch rpm from the lx zone
during the install process
6369493 zoneadm can't accept 'global' as a valid zone name.
6372605 SUNWlx delivers kernel modules in different locations
6373460 BrandZ lx_install use gtar for bzip2 and gz, but SUN tar for .tar,
GTAR must be use consistenly
6373984 SUNWlx needs to be split into two packages
6374606 ::nm -D without an object isn't working on a branded process
6375624 BrandZ must implement INSTALL mode for correct function of RPM
based install and update/upgrade
6378395 lx_mknod() is too restrictive of operations
6391195 clone() CLONE_THREAD fails to return created thread id to caller
6392193 'zoneadm list -v' displayed a messy BRAND in a native local zone.
6400694 The 'export' subcommand of zonecfg displayed wrong information for
linux zone.
6404260 'zoneadm clone' should verify that the target and source have the
same brand
6404323 if call 'zoneadm -z linuxzone install -d ' with a relative linux tar
ball path, error message is bad
6404830 zone console messages should be read through /proc/kmsg
6404996 The process hung during pthread testing on amd64.
6405286 Linux zones can't be cloned
6405901 Can't translate symbols in branded a.out
6409878 lx_send needs to support additional socket flags
6409878 lx_send needs to support additional socket flags (partial)
6415656 signal blocked in Child process will be 'automatically' released
before child exit
6420333 linux doesn't merge POLLOUT and POLLWRNORM
6420341 user signal handlers may be called with an incorrect %gs
6422040 SPECjbb2005 causes java to abort
6422515 lx_ptm should use I_STR for PTSOPENCB ioctl
6424886 Branded zones should support attach/detach sanely
6424887 Should support brand-specific verification in zoneadm and zonecfg
6424891 zone/brand interfaces need cleanup
6424897 zoneadm boot should be able to specify both -s and -f
6425903 "test" S10 brand should not be distributed
6426265 *BrandZ* Localization of lx_brand & lx_audio_cfg messages
6427459 Non-native zone creation is broken
6427624 blank lines in pldd output
6427732 pmap doesn't show meaningful paths for many objects
6427743 lx_elfexec should use global brand name definition
6428670 Attempt to execute 64-bit linux binary exposes global zone pathname
6430827 can not send signal to a set of child processes
6432365 Branded zone creation should be template-based
6434577 assertion failed: (brand_action != 1)
6434766 Brand templates are in the wrong place
6435859 linker code review changes
6435913 lx_librtld_db never succeeds in iterating over lx link map
6435914 lx_librtld_db in 64-bit processes fails to understand target data
objects
6436865 sendto() seg faults on bad socklen_t
6436959 lx_brand.so.1 should be more careful with its alloca() usage
6437563 msg_controllen == 0 is dumb but not EINVAL on Linux
6437977 ptools on more than one process warn about librtld_db helper
6438077 Need rudimentary /proc/net support
6438113 /proc loads wrong object for a.out when debugging core files
6439785 code review changes for syscall infrastructure
6439785 code review changes for syscall infrastructure
6440667 lx /proc code review changes
6441413 getpriority set errno to EPERM when it is passed invalid args
6442045 no /home dir built after installing centos zone, so useradd failed
6442428 My god, the futex(2) code is horrible.
6442592 pselect() failed because it cannot be allocated memory
6444059 move zone privilege lists into brand config file
6444065 library dependancies could still use some love
6444068 stat and friends could use some love
6444315 failure to load brand module will cause a system panic
6444612 stat() for block devices in linux returns 0 for st_size
6444631 lx_brand should let the kernel fail block/char mknod requests.
6444741 Miscellaneous code review changes needed for lx_brand library and
signals
6444793 64-bit crle config fails during s10 brand installation 32-bit systems
6444817 the brandz gate should be wx nits clean
6445225 brandsys() should be audited
6445253 uucopystr(2) should be known to libproc
6448193 need /proc/cpuinfo
6450776 Segmentation fault from init after install putback on Opteron systems
6450895 Installer scripts don't handle existing links properly
6451107 rpm's call to cpio may see a rename() failure with an errno of EBUSY
6451112 install log messages are TOO verbose
6451283 RedHat processes may core dump due to an error in %gs handling
6451474 Install log output could use a bit of cleanup
6451528 distro files are rife with unresolved dependencies
6451639 Installer should be more verbose during RPM installation
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss at opensolaris dot org
|
|
|
Posts:
1,539
From:
Cobourg Ontario Canada
Registered:
4/27/05
|
|
|
|
Re: Announcement: BrandZ build 43 now available
Posted:
Jul 22, 2006 3:04 PM
in response to: nilsn
|
|
> Greetings,
>
> It's been quite a while, but a new BrandZ build is now available. This
> build brings us into sync with nv43, fixes more than the usual number of
> bugs, and introduces our new installer.
>
> The new installer will continue to support tarballs, but we can finally
> install Linux zones directly from the CentOS 3.x install media - either CDs
> or ISO images. The final release will also support Red Hat 3.x, but the
> underlying mechanism is the same.
>
> This is likely to be our final release prior to integrating with Solaris,
> so we are getting pretty close to what we consider production-quality.
> There are still a number of bugs in here, but it's no longer something you
> need to be scared of.
wow .. lxrun is all grown up !
I will see if I can toss the Lotus Notes for Linux at it.
Dennis Clarke
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss at opensolaris dot org
|
|
|
|
|
