Thierry Parmentelat [Tue, 8 Dec 2015 10:19:19 +0000 (11:19 +0100)]
Setting tag bootmanager-5.3-3
patch for f23 as of dec. 2015 where kernel and initrd show up in
a new location under /boot
plus bugfix with format and single { and } for awk args
Thierry Parmentelat [Mon, 7 Dec 2015 15:18:59 +0000 (16:18 +0100)]
bugfix (would not find kernel at all)
+ factorized kernel and initrd
Thierry Parmentelat [Mon, 7 Dec 2015 13:01:37 +0000 (14:01 +0100)]
Merge branch 'master' of ssh://git.onelab.eu/git/bootmanager
* 'master' of ssh://git.onelab.eu/git/bootmanager:
Setting tag bootmanager-5.3-2 not a noarch package anymore (indeed it does come with binaries) fix ssl connection for runlevelagent for recent pythons for fedora23, ignore if rsa1 key generation fails
Thierry Parmentelat [Mon, 7 Dec 2015 13:01:28 +0000 (14:01 +0100)]
bugfixes for the location of kernel and initrd as observed on recent f23
plus, a couple of calls to format() had real {} that needed doubling
Thierry Parmentelat [Fri, 13 Nov 2015 11:56:21 +0000 (12:56 +0100)]
Setting tag bootmanager-5.3-2
not a noarch package anymore (indeed it does come with binaries)
fix ssl connection for runlevelagent for recent pythons
for fedora23, ignore if rsa1 key generation fails
Ciro Scognamiglio [Tue, 10 Nov 2015 13:29:41 +0000 (14:29 +0100)]
added check for ssl _create_unverified_context when talking to API
Thierry Parmentelat [Mon, 9 Nov 2015 11:38:57 +0000 (12:38 +0100)]
same as previous change : ignore error when producing the rsa1 ssh key
Thierry Parmentelat [Mon, 9 Nov 2015 10:40:06 +0000 (11:40 +0100)]
when generating keys inside the bootcd environment, ignore errors about the rsa1 key type, as support for this has been dropped apparently in f23
Thierry Parmentelat [Sat, 7 Nov 2015 13:00:59 +0000 (14:00 +0100)]
not noarch anymore
Thierry Parmentelat [Fri, 30 Oct 2015 15:33:14 +0000 (16:33 +0100)]
make debugmode can push this code on board a node in debug mode (aka safeboot)
Thierry Parmentelat [Thu, 8 Oct 2015 12:45:29 +0000 (14:45 +0200)]
fix for when upgrading too old nodes
Thierry Parmentelat [Fri, 26 Jun 2015 11:54:05 +0000 (13:54 +0200)]
Setting tag bootmanager-5.3-1
Use TLSv1 to connect to myplc, instead of SSLv3 that is known to be broken
Can implement 'upgrade' in addition to 'reinstall' boot state
upgrade does essentially the same as reinstall except that slices are preserved
upgrade works only on nodes already running containers
because /vservers/ needs to be a btrfs filesystem
pycurl is now a strong requirement (old curl-based code removed)
bugfix for ssh key generation (were all typed rsa1)
a lot of prettification
Thierry Parmentelat [Fri, 26 Jun 2015 11:35:47 +0000 (13:35 +0200)]
remove old code that was filling in for when pycurl is not available
even f12 bootCDs have pycurl
Thierry Parmentelat [Thu, 25 Jun 2015 21:49:08 +0000 (23:49 +0200)]
simplified PARTITIONS to use one entry per each partition
we used to have e.g. PARTITIONS['root'] and PARTITIONS['mapper-root']
but we don't need them both and it makes things confusing
Thierry Parmentelat [Thu, 25 Jun 2015 21:30:17 +0000 (23:30 +0200)]
vgchange is required so that /dev/mapper/* and /dev/planetlab/* show up when upgrading
Thierry Parmentelat [Thu, 25 Jun 2015 21:02:03 +0000 (23:02 +0200)]
add debug messages about the status of partitions
Thierry Parmentelat [Thu, 25 Jun 2015 16:29:52 +0000 (18:29 +0200)]
try to fix upgrade mode - run vgscan --mknodes
apparently not enough though
Thierry Parmentelat [Thu, 25 Jun 2015 16:29:15 +0000 (18:29 +0200)]
notes about an idea to rpm install a more recent btrfs-progs rpm in f14 boot-cds, but no place in the ramfs...
Thierry Parmentelat [Thu, 25 Jun 2015 11:09:08 +0000 (13:09 +0200)]
upgrade mode was not working; CheckForNewDisks was erroneously taking the disk as not yet in the volume, and would destroy everything
this is now taken care of by beinf a llittle more conservative in ChekForNewDisks
Thierry Parmentelat [Wed, 24 Jun 2015 14:47:45 +0000 (16:47 +0200)]
rename util get_block_device_list into get_block_devices_dict and clean up its usage
Thierry Parmentelat [Wed, 24 Jun 2015 14:44:12 +0000 (16:44 +0200)]
breakpoint messages always show up
Thierry Parmentelat [Wed, 24 Jun 2015 13:44:59 +0000 (15:44 +0200)]
use git ls-files to make tags
Thierry Parmentelat [Wed, 24 Jun 2015 12:59:08 +0000 (14:59 +0200)]
small cosmetic changes
Thierry Parmentelat [Tue, 23 Jun 2015 13:36:56 +0000 (15:36 +0200)]
prettyfied
Thierry Parmentelat [Tue, 23 Jun 2015 11:22:46 +0000 (13:22 +0200)]
a first stab at upgrading nodes
not working yet, but I need to build all this
Thierry Parmentelat [Tue, 23 Jun 2015 07:24:37 +0000 (09:24 +0200)]
always use print_function
Thierry Parmentelat [Sun, 21 Jun 2015 17:08:50 +0000 (19:08 +0200)]
use TLSv1 instead of SSLv3
Thierry Parmentelat [Wed, 29 Apr 2015 09:30:41 +0000 (11:30 +0200)]
remove breakpoints, and use log instead of print when possible
Thierry Parmentelat [Wed, 29 Apr 2015 09:25:49 +0000 (11:25 +0200)]
still prettifying
Thierry Parmentelat [Wed, 29 Apr 2015 09:03:03 +0000 (11:03 +0200)]
was meant with previous commit
Thierry Parmentelat [Tue, 28 Apr 2015 11:21:13 +0000 (13:21 +0200)]
oops
Thierry Parmentelat [Tue, 28 Apr 2015 11:09:34 +0000 (13:09 +0200)]
prettifying source/
comprehension instead of map
Thierry Parmentelat [Tue, 28 Apr 2015 10:15:28 +0000 (12:15 +0200)]
this well deserves a minor bump in version number
Thierry Parmentelat [Tue, 28 Apr 2015 10:14:42 +0000 (12:14 +0200)]
done prettifying
Thierry Parmentelat [Tue, 28 Apr 2015 09:53:40 +0000 (11:53 +0200)]
bugfix for ssh key generation (were all typed rsa1)
Thierry Parmentelat [Tue, 28 Apr 2015 09:53:12 +0000 (11:53 +0200)]
more prettifying
Thierry Parmentelat [Tue, 28 Apr 2015 09:21:39 +0000 (11:21 +0200)]
oops
Thierry Parmentelat [Tue, 28 Apr 2015 08:47:30 +0000 (10:47 +0200)]
oops
Thierry Parmentelat [Tue, 28 Apr 2015 00:15:48 +0000 (02:15 +0200)]
no real change, just made prettier with a more standard layout - half of steps
Thierry Parmentelat [Fri, 24 Apr 2015 15:15:27 +0000 (17:15 +0200)]
new boot_state 'upgrade' is recognized much like 'reinstall'
in this first rough implementation, a node in 'upgrade' mode goes through almost exactly the same steps as 'reinstall' except for re-partitioning
one possible drawback of this approach is we might en up with too much stuff in /
that is to say we might need to clean up manually some areas of the system to avoid accumulating things from previous releases
Thierry Parmentelat [Fri, 24 Apr 2015 15:08:03 +0000 (17:08 +0200)]
purely cosmetic: white spaces, .format rather than %, this kind of things only
Thierry Parmentelat [Fri, 24 Apr 2015 14:07:28 +0000 (16:07 +0200)]
make sync does not use lxc-enter-namespace
Thierry Parmentelat [Fri, 24 Apr 2015 13:16:32 +0000 (15:16 +0200)]
only cosmetic
Thierry Parmentelat [Wed, 16 Jul 2014 10:13:07 +0000 (12:13 +0200)]
Setting tag bootmanager-5.2-5
runs AnsibleHook, that optionnally runs playbooks (by default, usual behaviour)
introduces the ONE_PARTITION configuration variable
some suport for chainbooting ubuntu
Thierry Parmentelat [Thu, 15 May 2014 09:28:36 +0000 (11:28 +0200)]
Merge branch 'ansible'
Thierry Parmentelat [Mon, 28 Apr 2014 11:00:06 +0000 (13:00 +0200)]
Setting tag bootmanager-5.2-4
no functional change, only tweaks in Makefile for interating with test environment
Thierry Parmentelat [Wed, 16 Apr 2014 22:12:40 +0000 (00:12 +0200)]
define extensions in vars even if not yet set
Sapan Bhatia [Wed, 16 Apr 2014 06:41:41 +0000 (02:41 -0400)]
Merge branch 'ansible' of ssh://git.planet-lab.org/git/bootmanager into ansible
Sapan Bhatia [Wed, 16 Apr 2014 06:41:37 +0000 (02:41 -0400)]
The extensions variable is not expected to be in the configuration file if the latest version of MyPLC is installed
Thierry Parmentelat [Wed, 16 Apr 2014 06:24:15 +0000 (08:24 +0200)]
for debugging - print stack in case of exceptions caught at toplevel
Thierry Parmentelat [Wed, 16 Apr 2014 06:18:41 +0000 (08:18 +0200)]
fix make sync for the current state of the test infra
Thierry Parmentelat [Wed, 16 Apr 2014 06:18:41 +0000 (08:18 +0200)]
fix make sync for the current state of the test infra
Sapan Bhatia [Fri, 4 Apr 2014 21:38:43 +0000 (17:38 -0400)]
More workarounds for MyPLC 5.0
Sapan Bhatia [Fri, 4 Apr 2014 21:34:19 +0000 (17:34 -0400)]
Workaround to make extensions variable work with MyPLC 5.0
Sapan Bhatia [Fri, 4 Apr 2014 21:18:18 +0000 (17:18 -0400)]
Copy and paste error
Sapan Bhatia [Fri, 4 Apr 2014 20:58:41 +0000 (16:58 -0400)]
Add AnsibleHook step to list of available modules
Sapan Bhatia [Thu, 3 Apr 2014 18:46:30 +0000 (14:46 -0400)]
Set ONE_PARTITION if you want to make one giant root partition
Sapan Bhatia [Thu, 3 Apr 2014 18:39:30 +0000 (14:39 -0400)]
Changes to support running BootManager 5.2 on MyPLC < 5.2
Sapan Bhatia [Tue, 1 Apr 2014 12:43:13 +0000 (08:43 -0400)]
Step to hook into Ansible
Sapan Bhatia [Tue, 1 Apr 2014 04:08:04 +0000 (00:08 -0400)]
Ignore vservers partition
Sapan Bhatia [Tue, 1 Apr 2014 04:03:10 +0000 (00:03 -0400)]
Fixed path to Ubuntus initrd
Sapan Bhatia [Tue, 1 Apr 2014 03:55:29 +0000 (23:55 -0400)]
Look for kernel in image, not in root
Sapan Bhatia [Tue, 1 Apr 2014 03:48:13 +0000 (23:48 -0400)]
Drop node manager and node update
Sapan Bhatia [Tue, 1 Apr 2014 03:42:38 +0000 (23:42 -0400)]
Drop call to authconfig in generic distribution
Sapan Bhatia [Tue, 1 Apr 2014 03:37:02 +0000 (23:37 -0400)]
Don't write vservers to fstab for non-rpm
Sapan Bhatia [Tue, 1 Apr 2014 03:15:46 +0000 (23:15 -0400)]
Support non-rpm distributions
Sapan Bhatia [Tue, 1 Apr 2014 02:18:03 +0000 (22:18 -0400)]
Support kernel check other than rpm
Sapan Bhatia [Mon, 31 Mar 2014 23:15:19 +0000 (19:15 -0400)]
Tabulation issue
Sapan Bhatia [Mon, 31 Mar 2014 23:13:07 +0000 (19:13 -0400)]
Introduce new variable: ONE_PARTITION for indicating single-partition installations
Sapan Bhatia [Mon, 31 Mar 2014 23:03:38 +0000 (19:03 -0400)]
Typo
Sapan Bhatia [Mon, 31 Mar 2014 23:00:44 +0000 (19:00 -0400)]
Don't mount vservers partition if absent
Sapan Bhatia [Mon, 31 Mar 2014 22:42:34 +0000 (18:42 -0400)]
Don't mount vservers partition if absent
Sapan Bhatia [Mon, 31 Mar 2014 22:29:42 +0000 (18:29 -0400)]
ROOT_SIZE is a string not a number
Sapan Bhatia [Mon, 31 Mar 2014 21:31:01 +0000 (17:31 -0400)]
Identify kernel version in a way that does not depend on rpm
Sapan Bhatia [Mon, 31 Mar 2014 21:20:00 +0000 (17:20 -0400)]
Support having only one partition
Thierry Parmentelat [Tue, 25 Mar 2014 13:19:35 +0000 (14:19 +0100)]
Setting tag bootmanager-5.2-3
some old f18 bootCDs do not support mkfs.btrfs -f
so invoke this option only when supported
Thierry Parmentelat [Fri, 21 Mar 2014 17:19:02 +0000 (18:19 +0100)]
invoke mkfs.btrfs with -f only if supported
Thierry Parmentelat [Fri, 21 Mar 2014 15:12:59 +0000 (16:12 +0100)]
Setting tag bootmanager-5.2-2
conf_files.py is expected in /usr/share/NodeManager, not in /etc/init.d any more
smarter for locating initrd, for f20
add -f to mkfs.btrfs - sometimes hangs otherwise
Thierry Parmentelat [Tue, 14 Jan 2014 14:27:18 +0000 (15:27 +0100)]
typo
Thierry Parmentelat [Tue, 14 Jan 2014 14:27:02 +0000 (15:27 +0100)]
avoid using conf_files initscript that might go when moving to systemd
Thierry Parmentelat [Sat, 11 Jan 2014 15:14:48 +0000 (16:14 +0100)]
initrd has a new location under /boot/ in f20
Thierry Parmentelat [Sun, 22 Dec 2013 09:50:37 +0000 (10:50 +0100)]
Thomas's fix about calling mkfs.btrfs with the -f option
Thierry Parmentelat [Thu, 7 Mar 2013 07:39:09 +0000 (08:39 +0100)]
Setting tag bootmanager-5.2-1
merged the branches for vserver and lxc
requires the rest of 5.2 - notably 'virt' in GetNodeFlavour
note that WriteModprobeConfig and MakeInitrd are turned off for lxc nodes
also note that fsck management for btrfs/lxc is still weak
vs_ROOT_SIZE=14G lxc_ROOT_SIZE=70G
vs_TOTAL_MINIMUM_DISK_SIZE=50G lxc_TOTAL_MINIMUM_DISK_SIZE=120G
expects ntpd to be turned on in the nodeimage
Thierry Parmentelat [Wed, 27 Feb 2013 16:18:38 +0000 (17:18 +0100)]
turning on ntpd is now done in the nodeimage (post script) directly
Thierry Parmentelat [Wed, 27 Feb 2013 12:41:56 +0000 (13:41 +0100)]
bugfixes
Thierry Parmentelat [Wed, 27 Feb 2013 12:41:33 +0000 (13:41 +0100)]
merged 5.0 (traditional vserver-based) and 5.1 (aka lxc_devel)
now retrieves GetNodeFlavour as part of GetNodeDetails
and stores result as global vars['node_flavour']
this one now is expected to include the 'virt' tag
as either 'vs' or 'lxc' and act accordingly
STILL MISSING
fsck under btrfs, i.e.
* turn off periodic fsck in InstallPartitionDisk, and
* explicitly run fsck in ValidateNodeInstall
ALSO
the MakeInitrd and WriteModprobeConfig steps are turned off with lxc
Thierry Parmentelat [Wed, 27 Feb 2013 06:40:12 +0000 (07:40 +0100)]
bump major to 5.2 - preparing for the merge
Thierry Parmentelat [Tue, 26 Feb 2013 20:13:16 +0000 (21:13 +0100)]
log more details
Thierry Parmentelat [Tue, 26 Feb 2013 16:14:17 +0000 (17:14 +0100)]
just print out node's 'virt' as gotten from GetNodeFlavour
Thierry Parmentelat [Fri, 31 Aug 2012 17:40:15 +0000 (19:40 +0200)]
Setting tag bootmanager-5.0-24
run parted with --script to avoid it to hang
Thierry Parmentelat [Fri, 31 Aug 2012 09:58:21 +0000 (11:58 +0200)]
run parted with the --script option to prevent from hanging
also use more explicit --list option instead of -l
Thierry Parmentelat [Mon, 9 Jul 2012 09:44:32 +0000 (11:44 +0200)]
Setting tag bootmanager-5.0-23
added support for disks larger than 2Tb using gpt instead of msdos
Thierry Parmentelat [Sat, 30 Jun 2012 06:48:48 +0000 (08:48 +0200)]
improve rendering in log
Thierry Parmentelat [Fri, 29 Jun 2012 18:19:48 +0000 (20:19 +0200)]
another angle to the 'try msdos, fallback to gpt' approach
Thierry Parmentelat [Fri, 29 Jun 2012 16:35:19 +0000 (18:35 +0200)]
add a comment about why this partitioning code is still broken
Sapan Bhatia [Thu, 28 Jun 2012 08:25:43 +0000 (04:25 -0400)]
sfdisk does not support gpt partition tables
Thierry Parmentelat [Wed, 27 Jun 2012 11:27:05 +0000 (13:27 +0200)]
partition tables using gpt instead of msdos with disks larger than 2TB
Thierry Parmentelat [Tue, 15 May 2012 11:16:11 +0000 (13:16 +0200)]
Setting tag bootmanager-5.0-22
bootmanager log clearly states duration of download and extraction of node image
Thierry Parmentelat [Tue, 15 May 2012 10:20:34 +0000 (12:20 +0200)]
whoops
Thierry Parmentelat [Tue, 15 May 2012 08:27:57 +0000 (10:27 +0200)]
easier to retrieve duration for downloading/extracting install image