Thierry Parmentelat [Thu, 31 Mar 2016 08:12:48 +0000 (10:12 +0200)]
cosmetic
Thierry Parmentelat [Thu, 31 Mar 2016 07:51:27 +0000 (09:51 +0200)]
2to3 -f except
Thierry Parmentelat [Thu, 31 Mar 2016 07:49:49 +0000 (09:49 +0200)]
xmlrpclib.Fault for py3
Thierry Parmentelat [Thu, 31 Mar 2016 07:27:01 +0000 (09:27 +0200)]
a lingering raise
Thierry Parmentelat [Thu, 31 Mar 2016 06:57:04 +0000 (08:57 +0200)]
2to3 -f raise
Thierry Parmentelat [Thu, 31 Mar 2016 06:55:05 +0000 (08:55 +0200)]
2to3 -f print
from __future__ import print_function
Thierry Parmentelat [Thu, 31 Mar 2016 06:41:55 +0000 (08:41 +0200)]
dos2unix'ed
Thierry Parmentelat [Wed, 24 Feb 2016 09:41:54 +0000 (10:41 +0100)]
Applied patch from GENI by Aaron Helsinger
https://github.com/GENI-NSF/geni-tools/pull/894/files
Thierry Parmentelat [Fri, 5 Feb 2016 16:32:33 +0000 (17:32 +0100)]
sfi.py closer to PEP-8
Thierry Parmentelat [Fri, 5 Feb 2016 16:27:02 +0000 (17:27 +0100)]
sfi myslice more robust - won't fail when one slice is problematic
Thierry Parmentelat [Tue, 12 Jan 2016 09:13:04 +0000 (10:13 +0100)]
bugfix - issue seen on PLE with slice onelab.upmc.fit_demo.wireless_fit
there was a remaining underscore in the login_base produced by
hash_loginbase
Thierry Parmentelat [Fri, 8 Jan 2016 12:06:17 +0000 (13:06 +0100)]
sfi introspect
Thierry Parmentelat [Thu, 7 Jan 2016 09:09:34 +0000 (10:09 +0100)]
build geni_rspec_version in a consistent way
Thierry Parmentelat [Thu, 7 Jan 2016 09:09:06 +0000 (10:09 +0100)]
review all commands for arg count check
Thierry Parmentelat [Mon, 4 Jan 2016 10:16:17 +0000 (11:16 +0100)]
rename get_m2_pkey into get_m2_pubkey in unit test as well
fsaintma [Fri, 18 Dec 2015 12:41:16 +0000 (13:41 +0100)]
Merge branch 'geni-v3' of ssh://git.onelab.eu/git/sfa into geni-v3
fsaintma [Fri, 18 Dec 2015 12:40:54 +0000 (13:40 +0100)]
add Dead state for unvailable nodes
fsaintma [Fri, 18 Dec 2015 12:38:25 +0000 (13:38 +0100)]
improve get_reserve_nodes
Thierry Parmentelat [Thu, 17 Dec 2015 13:10:44 +0000 (14:10 +0100)]
Setting tag sfa-3.1-20
minor fixes for migrating on fedora23
Thierry Parmentelat [Thu, 17 Dec 2015 10:44:02 +0000 (11:44 +0100)]
bugfix - m2crypto verification function returns
. 1 in case of success
. but otherwise it may be 0 or -1
Thierry Parmentelat [Thu, 17 Dec 2015 10:38:05 +0000 (11:38 +0100)]
rename ambiguous pkey into pubkey
Thierry Parmentelat [Thu, 17 Dec 2015 10:31:23 +0000 (11:31 +0100)]
cosmetic changes only
Thierry Parmentelat [Wed, 16 Dec 2015 08:22:06 +0000 (09:22 +0100)]
better patch for unverified_context
Thierry Parmentelat [Thu, 10 Dec 2015 10:56:18 +0000 (11:56 +0100)]
prettified speaksfor a little
Thierry Parmentelat [Thu, 10 Dec 2015 10:47:53 +0000 (11:47 +0100)]
bugfix, correct previous change on locating xmlsec1
Thierry Parmentelat [Thu, 10 Dec 2015 09:17:37 +0000 (10:17 +0100)]
no change, just sfi.py prettyfied wrt spaces and assignments
Thierry Parmentelat [Thu, 10 Dec 2015 08:57:35 +0000 (09:57 +0100)]
more aggressively check for xmlsec1 being installed,
raise an exception otherwise
Thierry Parmentelat [Tue, 8 Dec 2015 10:29:29 +0000 (11:29 +0100)]
Setting tag sfa-3.1-19
imported changes from GENI as reported - Aaron Helsinger
minimal changes so that parts can be imported from nepi/py3
iotlab driver : fix ASAP jobs with state != Waiting, Running - Frederic
sfi client more accurately advertises rspec version - Loic
+ bugfix in initscript
Thierry Parmentelat [Fri, 4 Dec 2015 11:03:56 +0000 (12:03 +0100)]
blind attempt to undo renaming of
credential.type into credential.cred_type
Thierry Parmentelat [Thu, 26 Nov 2015 08:43:23 +0000 (09:43 +0100)]
changes proposed by Aaron Helsinger and GENI
Thierry Parmentelat [Thu, 8 Oct 2015 15:32:43 +0000 (17:32 +0200)]
minimal changes to have this imported by nepi in py3
Loic Baron [Mon, 28 Sep 2015 14:01:27 +0000 (16:01 +0200)]
Merge branch 'geni-v3' of ssh://git.onelab.eu/git/sfa into geni-v3
Loic Baron [Mon, 28 Sep 2015 14:00:20 +0000 (16:00 +0200)]
SFI client passing -r parameter even if the rspec_type is not known by sfi.py client
Thierry Parmentelat [Wed, 22 Jul 2015 09:57:23 +0000 (11:57 +0200)]
use print_function in sfi.py
Thierry Parmentelat [Wed, 22 Jul 2015 09:53:11 +0000 (11:53 +0200)]
fix broken sfi config
Thierry Parmentelat [Tue, 21 Jul 2015 12:27:25 +0000 (14:27 +0200)]
slight change in the initscript
this is all very odd;
out of the blue the SFA service within its f20 VM (inside bigjohn that is a f21 LXC host) was not starting properly any longer..
the test to decide whether to start postgresql had to be rewritten, reading now
check || { start-db ; ... }
rather than
if [ ! check ] ; then start-db ... ; fi
fsaintma [Thu, 11 Jun 2015 07:24:10 +0000 (09:24 +0200)]
fix ASAP jobs with state != Waiting,Running
fsaintma [Wed, 10 Jun 2015 07:23:47 +0000 (09:23 +0200)]
Merge branch 'geni-v3' of ssh://git.onelab.eu/git/sfa into geni-v3
fsaintma [Wed, 10 Jun 2015 07:23:32 +0000 (09:23 +0200)]
fix ASAP duration
Thierry Parmentelat [Mon, 8 Jun 2015 10:24:45 +0000 (12:24 +0200)]
Setting tag sfa-3.1-18
incorporated Frederic Saint Marcel's addition of ASAP management tag
fsaintma [Mon, 8 Jun 2015 08:41:14 +0000 (10:41 +0200)]
Merge branch 'geni-v3' of ssh://git.onelab.eu/git/sfa into geni-v3
fsaintma [Mon, 8 Jun 2015 08:41:11 +0000 (10:41 +0200)]
add ASAP mgt and fix start_time with ASAP jobs
Thierry Parmentelat [Fri, 5 Jun 2015 12:00:01 +0000 (14:00 +0200)]
Setting tag sfa-3.1-17
workaround for 'name' not being exposed properly by List() on authority objects
fix a corner case in PL importer
trashed module registry_manager_openstack
Thierry Parmentelat [Fri, 5 Jun 2015 11:57:41 +0000 (13:57 +0200)]
workaround the mysterious issue with sqlalchemy that returns instances of RegAuthority that do not
have their __dict__ in line with the object's contents (specifically missing 'name')
a possibly other instance of that same issue was worked around in Update() using an extra print statement,
which I prefer to leave as is for now as I cannot test that thoroughly, but a similar approach would probably
be a solution in this case as well
Thierry Parmentelat [Fri, 5 Jun 2015 11:17:08 +0000 (13:17 +0200)]
cosmetic
Thierry Parmentelat [Fri, 5 Jun 2015 10:43:22 +0000 (12:43 +0200)]
cosmetic
Thierry Parmentelat [Fri, 5 Jun 2015 10:43:06 +0000 (12:43 +0200)]
prettyfied registry manager (+format)
Thierry Parmentelat [Fri, 5 Jun 2015 10:40:15 +0000 (12:40 +0200)]
with instead of open/close
Thierry Parmentelat [Fri, 5 Jun 2015 10:06:23 +0000 (12:06 +0200)]
Record.todict behaves suspisciously
no change for now, but we disambiguate between XML.todict() and Record.todict()
so we can more easily spot which is which
Thierry Parmentelat [Fri, 5 Jun 2015 10:03:23 +0000 (12:03 +0200)]
cleaned up saving-to-file functions in sfi - now always displays a message when writing a file
Thierry Parmentelat [Fri, 5 Jun 2015 10:02:03 +0000 (12:02 +0200)]
trashed registry_manager_openstack
Thierry Parmentelat [Fri, 5 Jun 2015 08:10:29 +0000 (10:10 +0200)]
remove printing 'name' in low-level record utilities, as UTF-8 values can create serious errors on the server side if ill-configured
Thierry Parmentelat [Fri, 5 Jun 2015 08:09:31 +0000 (10:09 +0200)]
fix corner case in PL importer
Thierry Parmentelat [Fri, 5 Jun 2015 07:41:21 +0000 (09:41 +0200)]
all sfi commands that support -t/--type now accept a 2-chars shorthand
Thierry Parmentelat [Fri, 5 Jun 2015 07:14:28 +0000 (09:14 +0200)]
add a useless print statement to work around a weird issue we have with sqlalchemy objects and their __dict__ value being wrong
this is the second occurrence of that issue and a better solution is clearly needed
in the mix, record.todict() now expects a tuple of types to exclude; this way we can use isinstance natively
Thierry Parmentelat [Fri, 5 Jun 2015 07:12:08 +0000 (09:12 +0200)]
repr() of Reg objects shows email and name if present
Thierry Parmentelat [Fri, 5 Jun 2015 07:11:27 +0000 (09:11 +0200)]
no change - spaces
Thierry Parmentelat [Thu, 4 Jun 2015 12:18:29 +0000 (14:18 +0200)]
Setting tag sfa-3.1-16
added a new builtin column 'name' for authorities in the sfa registry
this is kept in sync with MyPLC's site names when relevant
sfa update -t authority thus now has a new -n/--name option
sfi register or update can specify record type on only 2 characters (au, us, no, or sl)
reviewed Describe and Allocate wrt slice tags for a PL AM:
Describe now exposes all slice tags with a 'scope' being 'sliver' or 'slice'
Allocate now by default ignores incoming slice tags
Allocate's options can mention 'pltags' among 'ignore', 'append', 'sync'
default being 'ignore'
in 'ignore' mode, slice tags are unchanged in the PL db
in 'append' mode, slice tags from the rspec are added to the db unless
they are already present
in 'sync' mode, the code attempts to leave the PL db in sync with the tags
provided in rspec; this can be dangerous and is thus no longer the default
behaviour
Thierry Parmentelat [Thu, 4 Jun 2015 12:07:14 +0000 (14:07 +0200)]
user needs to specify 'pltags':'sync' in the options to Allocate to use standard behaviour, insted of more confusing 'match'
Thierry Parmentelat [Wed, 3 Jun 2015 15:54:17 +0000 (17:54 +0200)]
removed the 'append' option to Allocate in the PL driver;
replaced with the 'pltags' option that can be either 'ignore' (default), 'append' (only add the provided tags if not yet present) or anything else (try to do the magic, which is known to be a little fragile)
Thierry Parmentelat [Wed, 3 Jun 2015 15:12:37 +0000 (17:12 +0200)]
add a docstring note about Allocate and the append option
Thierry Parmentelat [Wed, 3 Jun 2015 14:17:43 +0000 (16:17 +0200)]
major fix about slice tags
no slice tags was exposed by sfi describe before this change
we now expose them properly with a 'scope' field that can be either 'slice' or 'sliver'
note that nodegroup-based tags are still silently ignored at this point
---
using allocate with the 'append' option is probably the best approach anyway, with an empty list of slice tags; this avoid resetting slice tags
Thierry Parmentelat [Wed, 3 Jun 2015 14:15:18 +0000 (16:15 +0200)]
improved sfi describe : show help is wrong number of args, use a context manager for opening input file
Thierry Parmentelat [Wed, 3 Jun 2015 14:14:38 +0000 (16:14 +0200)]
no change - various tweaks here and there
Thierry Parmentelat [Wed, 3 Jun 2015 10:20:32 +0000 (12:20 +0200)]
prettyfied - spaces
Thierry Parmentelat [Wed, 3 Jun 2015 10:13:28 +0000 (12:13 +0200)]
added debug traces
Thierry Parmentelat [Wed, 3 Jun 2015 10:12:44 +0000 (12:12 +0200)]
no change - use format instead of % in plimporter pldriver plslices
Thierry Parmentelat [Wed, 3 Jun 2015 09:54:53 +0000 (11:54 +0200)]
belongs with previous change
Thierry Parmentelat [Wed, 3 Jun 2015 09:51:36 +0000 (11:51 +0200)]
planetlab import : site name is authorititative over authority's name
Thierry Parmentelat [Wed, 3 Jun 2015 09:50:43 +0000 (11:50 +0200)]
support for e.g. sfi discover help
Thierry Parmentelat [Wed, 3 Jun 2015 09:50:22 +0000 (11:50 +0200)]
no change - prettyfied sfi.py (uses format instead of %)
Thierry Parmentelat [Tue, 2 Jun 2015 15:40:16 +0000 (17:40 +0200)]
sfi describe/resources shows its default for -r in the help message
also we now have a new 'discover' alias for sfi resources
Thierry Parmentelat [Tue, 2 Jun 2015 15:13:38 +0000 (17:13 +0200)]
cleanup in plslices.verify_slice_attributes
no semantic change though
Thierry Parmentelat [Tue, 2 Jun 2015 15:12:49 +0000 (17:12 +0200)]
pgv2 - fix in the logic for retrieving slice attributes
Thierry Parmentelat [Tue, 2 Jun 2015 15:08:22 +0000 (17:08 +0200)]
document the options available to Allocate
Thierry Parmentelat [Tue, 2 Jun 2015 15:08:04 +0000 (17:08 +0200)]
minor tweak in rspec embedded tester - expects an xml filename, not a xml string
Thierry Parmentelat [Tue, 2 Jun 2015 15:07:17 +0000 (17:07 +0200)]
minor tweaks in version manager embedded tester
Thierry Parmentelat [Tue, 2 Jun 2015 15:06:48 +0000 (17:06 +0200)]
no change - prettyfied
Thierry Parmentelat [Tue, 2 Jun 2015 13:40:49 +0000 (15:40 +0200)]
again, this change belongs with the addition of the builtin 'name' column for authorities
Thierry Parmentelat [Tue, 2 Jun 2015 12:19:44 +0000 (14:19 +0200)]
this belongs with previous change on authority's name
this is how sfi update now does the right thing with both users and authority wrt email and name
Thierry Parmentelat [Tue, 2 Jun 2015 12:16:55 +0000 (14:16 +0200)]
sfi register and sfi update require a -t option; this now accepts 2-letter shorthands like -t us or -t no
as a matter of fact only the 2 first letters are considered
Thierry Parmentelat [Tue, 2 Jun 2015 12:15:51 +0000 (14:15 +0200)]
sfi list -v will now show user's email and authority's name as well
Thierry Parmentelat [Tue, 2 Jun 2015 12:15:05 +0000 (14:15 +0200)]
no change = prettyfied wrt spaces, commas and the like
Thierry Parmentelat [Tue, 2 Jun 2015 12:09:26 +0000 (14:09 +0200)]
authorities can also have a 'name' for standalone deployments
comes with a db migration to that effect as well
authority name can be set in sfi register and/or sfi update
in the mix, bugfix for user emails: sfi update ... --email foo
was not making it to the sfa db
fsaintma [Fri, 22 May 2015 09:02:11 +0000 (11:02 +0200)]
fix pylint error
fsaintma [Fri, 22 May 2015 08:57:20 +0000 (10:57 +0200)]
add information in Iot-LAB rspec node
fsaintma [Tue, 12 May 2015 11:50:02 +0000 (13:50 +0200)]
fix bug with lease duration in seconds
Thierry Parmentelat [Thu, 23 Apr 2015 08:53:37 +0000 (10:53 +0200)]
Setting tag sfa-3.1-15
major rework of the iotlab driver, that uses an IoT-lab REST API
and so does not need to interact with LDAP and OAR directly
deprecated cortexlab driver altogether
cosmetic changes in displaying credentials, rights and certificates
for hopefully more readable error messages
always start postgresql if not running (ignore /etc/myplc-release)
does not need lxc=enter-namespace anymore for make sync
Thierry Parmentelat [Thu, 16 Apr 2015 11:03:11 +0000 (13:03 +0200)]
do not rely on /etc/myplc-release to start or stop postgresql
instead: start it when not yet running, and record that in /etc/sfa/postgresql-started so we can stop it properly
Thierry Parmentelat [Thu, 16 Apr 2015 10:56:35 +0000 (12:56 +0200)]
cleanup startup script - no change
Thierry Parmentelat [Tue, 14 Apr 2015 15:02:28 +0000 (17:02 +0200)]
hopefully nicer and more helpful messages when verify_parent fails
exception should be shorter and easier to read, while full details still go in the logs
Thierry Parmentelat [Tue, 14 Apr 2015 15:01:45 +0000 (17:01 +0200)]
issue a warning when get_slivers is used against an existing but empty slice
e.g. with this code one gets an exception when trying to renew an existing slice that has no node
the warning would help spotting this
Thierry Parmentelat [Tue, 14 Apr 2015 11:00:07 +0000 (13:00 +0200)]
disambiguate get_subject and pretty_subject
Thierry Parmentelat [Tue, 14 Apr 2015 10:59:33 +0000 (12:59 +0200)]
lower noise due to alchemy session -> debug instead of info - ctd
Thierry Parmentelat [Tue, 14 Apr 2015 10:47:06 +0000 (12:47 +0200)]
more compact, and hopefully nicer versions of pretty_cred and pretty_cert
Thierry Parmentelat [Tue, 14 Apr 2015 10:32:13 +0000 (12:32 +0200)]
lower noise due to alchemy session -> debug instead of info
Thierry Parmentelat [Tue, 14 Apr 2015 10:01:57 +0000 (12:01 +0200)]
rename the 'cert' attribute of Certificate into 'x509'
Thierry Parmentelat [Tue, 14 Apr 2015 09:43:07 +0000 (11:43 +0200)]
do not use broken lxc-enter-namespace
Thierry Parmentelat [Tue, 14 Apr 2015 09:33:04 +0000 (11:33 +0200)]
renamed get_printable_subject into pretty_cert
Thierry Parmentelat [Tue, 14 Apr 2015 09:29:20 +0000 (11:29 +0200)]
tweaked verify_parent for more legible
add error in log with details on the comparison when raising "Delegated credential {} not signed by parent {}'s caller"