Thierry Parmentelat [Mon, 13 Apr 2015 09:40:35 +0000 (11:40 +0200)]
convenience macros restart-node and wait-node
Thierry Parmentelat [Mon, 13 Apr 2015 09:40:05 +0000 (11:40 +0200)]
get bootcd to run in dry-mode
Thierry Parmentelat [Mon, 13 Apr 2015 07:13:44 +0000 (09:13 +0200)]
miscell cosmetic + pass the substrate object to TestBonding so it can compute an available IP for a bonding node
Thierry Parmentelat [Sun, 12 Apr 2015 15:47:58 +0000 (17:47 +0200)]
cosmetic
Thierry Parmentelat [Sun, 12 Apr 2015 15:47:34 +0000 (17:47 +0200)]
bootmedium options and node tags in the config
Thierry Parmentelat [Sun, 12 Apr 2015 10:43:31 +0000 (12:43 +0200)]
bonding: exclude drupal when updating
Thierry Parmentelat [Sun, 12 Apr 2015 09:28:13 +0000 (11:28 +0200)]
more IPs for vplc
Thierry Parmentelat [Sat, 11 Apr 2015 18:12:49 +0000 (20:12 +0200)]
somehow the just-created bridge interface was found down on our qemu host...
plus, ip route add default won't work any more and I have to specify 0.0.0.0/0
Thierry Parmentelat [Sat, 11 Apr 2015 16:25:35 +0000 (18:25 +0200)]
keep stdin open in the ssh leg between host and guest
Thierry Parmentelat [Sat, 11 Apr 2015 11:14:10 +0000 (13:14 +0200)]
no more lxc-enter-namespace - use ssh
Thierry Parmentelat [Tue, 7 Apr 2015 09:30:49 +0000 (11:30 +0200)]
remove spurrious msgs from stderr
all classes in Substrate have a __repr__ for dbg
Thierry Parmentelat [Sat, 4 Apr 2015 14:03:45 +0000 (16:03 +0200)]
boxtops can stand the heat
Thierry Parmentelat [Fri, 3 Apr 2015 15:32:46 +0000 (17:32 +0200)]
nicer
Thierry Parmentelat [Fri, 3 Apr 2015 12:24:02 +0000 (14:24 +0200)]
needed for creating leases in python3
Thierry Parmentelat [Fri, 3 Apr 2015 08:57:21 +0000 (10:57 +0200)]
Setting tag tests-6.0-1
runs python3
mostly current infra is based on f21
known pending issue as we now have build VMs and vplcs in the same box
and Substrate needs a few tweaks to handle that more accurately
Thierry Parmentelat [Fri, 3 Apr 2015 08:55:54 +0000 (10:55 +0200)]
bump to version 6 as we now use python3
Thierry Parmentelat [Thu, 2 Apr 2015 16:18:57 +0000 (18:18 +0200)]
more smartly sense for vplc instances now that we run build VMs and vplc's in the same physical host
Thierry Parmentelat [Thu, 2 Apr 2015 16:07:55 +0000 (18:07 +0200)]
move vplc instances to buzzcocks - that is also the build box
Thierry Parmentelat [Thu, 2 Apr 2015 16:07:29 +0000 (18:07 +0200)]
run lxc-enter-namespace with --noseclabel
Thierry Parmentelat [Thu, 19 Mar 2015 20:24:23 +0000 (21:24 +0100)]
replace os.system with subprocess.call - add a timeout option in the mix
today for the first time ever a slice was not answering ssh but the ssh client would hang forever..
Thierry Parmentelat [Thu, 19 Mar 2015 20:21:24 +0000 (21:21 +0100)]
nicer
Thierry Parmentelat [Thu, 19 Mar 2015 20:16:53 +0000 (21:16 +0100)]
keep track of what has moved to py3
Thierry Parmentelat [Thu, 19 Mar 2015 10:24:29 +0000 (11:24 +0100)]
oops
Thierry Parmentelat [Wed, 18 Mar 2015 16:27:02 +0000 (17:27 +0100)]
SyntaxError: from __future__ imports must occur at the beginning of the file
Thierry Parmentelat [Wed, 18 Mar 2015 15:40:31 +0000 (16:40 +0100)]
revert tcptest to python2 as this runs in the slice context
use print() as per print_function though
Thierry Parmentelat [Wed, 18 Mar 2015 13:16:18 +0000 (14:16 +0100)]
need to invoke python3
Thierry Parmentelat [Wed, 18 Mar 2015 10:47:11 +0000 (11:47 +0100)]
python3 - first draft looks kind of OK
Thierry Parmentelat [Wed, 18 Mar 2015 09:36:51 +0000 (10:36 +0100)]
Setting tag tests-5.3-10
no functional change, but hopefully last version for python2
Thierry Parmentelat [Tue, 17 Mar 2015 20:25:35 +0000 (21:25 +0100)]
oops
Thierry Parmentelat [Tue, 17 Mar 2015 16:58:59 +0000 (17:58 +0100)]
one layer of polish
Thierry Parmentelat [Tue, 17 Mar 2015 09:49:39 +0000 (10:49 +0100)]
a first rough implementation for testing 'bonding' myplc's
typicallly, you would have 2 complete builds b1 and b2
then on testmaster in b1 you can do
$ run --bonding b2
or simply
$ run -g b2
in order to expose the nodeflavour and slicefamily provided by b2 to build b1
for now it only does the partial-rpms & yum install bootcd nodeimage noderepo & packages reindexing
need some more work to bring a actual nodes and slices online
Thierry Parmentelat [Mon, 16 Mar 2015 21:57:17 +0000 (22:57 +0100)]
ArgumentParser instead of OptionParser
Thierry Parmentelat [Mon, 16 Mar 2015 21:06:55 +0000 (22:06 +0100)]
more pretty-printing
Thierry Parmentelat [Mon, 16 Mar 2015 16:08:53 +0000 (17:08 +0100)]
no change - pretty printed mostly
Thierry Parmentelat [Thu, 12 Mar 2015 10:58:18 +0000 (11:58 +0100)]
release reed from its build duties
Thierry Parmentelat [Tue, 10 Mar 2015 10:23:44 +0000 (11:23 +0100)]
check_initscripts no longer a warning
Thierry Parmentelat [Mon, 9 Mar 2015 14:08:56 +0000 (15:08 +0100)]
make cross-check-tcp more robust : we need to wait for network presence on both ends of the link
Thierry Parmentelat [Sun, 8 Mar 2015 23:15:58 +0000 (00:15 +0100)]
have iterate-ping-node display its logdir
Thierry Parmentelat [Fri, 6 Mar 2015 23:09:05 +0000 (00:09 +0100)]
iterate-ping-node with the -o option
Thierry Parmentelat [Fri, 6 Mar 2015 10:12:37 +0000 (11:12 +0100)]
a utility script for hammering on the ping-node step
Thierry Parmentelat [Thu, 5 Mar 2015 12:43:59 +0000 (13:43 +0100)]
cosmetic - when displaying Completer short timeouts (in the order of seconds and not minutes)
Thierry Parmentelat [Thu, 5 Mar 2015 12:43:18 +0000 (13:43 +0100)]
make tcptest ready more stringent : check that eth0 indeed has an ipv4 address
Thierry Parmentelat [Tue, 3 Mar 2015 23:36:21 +0000 (00:36 +0100)]
rework check-tcp so that we first wait for the network to be ready in the sliver
Thierry Parmentelat [Tue, 3 Mar 2015 15:01:04 +0000 (16:01 +0100)]
the kernel flag systemd.log_level=debug is not set by default anymore, we add the option to GetBootMedium that does this
Thierry Parmentelat [Tue, 3 Mar 2015 13:51:04 +0000 (14:51 +0100)]
improve the output of tcp{server,client}
make sure this can be used next time this step fails
Thierry Parmentelat [Tue, 3 Mar 2015 13:50:07 +0000 (14:50 +0100)]
use consitent convention for displaying step names
Thierry Parmentelat [Tue, 3 Mar 2015 13:21:36 +0000 (14:21 +0100)]
use standard python presentation convention
Thierry Parmentelat [Tue, 3 Mar 2015 10:37:40 +0000 (11:37 +0100)]
show network status inside tcptest.py in server mode
Thierry Parmentelat [Tue, 3 Mar 2015 10:37:34 +0000 (11:37 +0100)]
cosmetic changes only
+ a comment on the occasional issue with cross-check-tcp
Thierry Parmentelat [Mon, 2 Mar 2015 20:18:59 +0000 (21:18 +0100)]
no need to wait that long for ping_node, it works very fast, or does not
Thierry Parmentelat [Mon, 2 Mar 2015 14:00:07 +0000 (15:00 +0100)]
comment for keeping track of the memory size
Thierry Parmentelat [Mon, 2 Mar 2015 13:58:37 +0000 (14:58 +0100)]
increase memory of test qemu VMs to 2Gb
Thierry Parmentelat [Mon, 2 Mar 2015 11:49:10 +0000 (12:49 +0100)]
a new config that tweaks the config and removes initscript-related annotations
Thierry Parmentelat [Sun, 1 Mar 2015 15:10:43 +0000 (16:10 +0100)]
pass dry_run to the ssh-node-{debug,boot} steps
Thierry Parmentelat [Sun, 1 Mar 2015 07:22:55 +0000 (08:22 +0100)]
nicer displays of various durations
Thierry Parmentelat [Sat, 28 Feb 2015 13:31:51 +0000 (14:31 +0100)]
welcome to buzzcocks
Thierry Parmentelat [Sat, 28 Feb 2015 11:19:08 +0000 (12:19 +0100)]
successful Completer prints duration
Thierry Parmentelat [Sat, 28 Feb 2015 09:59:28 +0000 (10:59 +0100)]
not sure if it's the 21 nodes that have a larger payload, or if it's because we have all kvm nodes in a single box now,
but we need to alllow for more time before the node pings
Thierry Parmentelat [Fri, 27 Feb 2015 22:17:33 +0000 (23:17 +0100)]
need to see how far the f21 build can go
Thierry Parmentelat [Fri, 27 Feb 2015 08:42:09 +0000 (09:42 +0100)]
bridge initialization on a f21 testbox; only uses ip, i.e. no ifconfig or route anymore
Thierry Parmentelat [Thu, 26 Feb 2015 15:09:44 +0000 (16:09 +0100)]
fail on purpose while all slices are running
Thierry Parmentelat [Wed, 25 Feb 2015 15:34:50 +0000 (16:34 +0100)]
move to another box
Thierry Parmentelat [Tue, 24 Feb 2015 16:50:24 +0000 (17:50 +0100)]
Merge branch 'sfa-geni-v3' of ssh://git.onelab.eu/git/tests into sfa-geni-v3
Thierry Parmentelat [Tue, 24 Feb 2015 16:50:00 +0000 (17:50 +0100)]
use a single recent test box instead of n small old boxes
Thierry Parmentelat [Wed, 18 Feb 2015 11:48:48 +0000 (12:48 +0100)]
Setting tag tests-5.3-9
add tests for sfa and empty rspecs
Thierry Parmentelat [Wed, 18 Feb 2015 09:07:25 +0000 (10:07 +0100)]
turn off the 'fail' step that was causing the builds to fail for forensics purposes
Thierry Parmentelat [Sun, 8 Feb 2015 11:13:54 +0000 (12:13 +0100)]
force failure after slice-fs-deleted
Thierry Parmentelat [Fri, 6 Feb 2015 21:43:32 +0000 (22:43 +0100)]
do not ignore slice-fs-deleted
Thierry Parmentelat [Tue, 14 Oct 2014 07:20:22 +0000 (09:20 +0200)]
estran out
Thierry Parmentelat [Mon, 22 Sep 2014 14:30:03 +0000 (16:30 +0200)]
estran : max vnodes 3->2, too small disk
Thierry Parmentelat [Fri, 22 Aug 2014 10:46:25 +0000 (12:46 +0200)]
decidedly... sfa_rspec_empty needs the sfi/pg dir to have been created beforehand
Thierry Parmentelat [Thu, 21 Aug 2014 16:28:21 +0000 (18:28 +0200)]
the sfa_{allocate,provision,check_slice_plc}_empty sequence needs to be done later
Thierry Parmentelat [Wed, 20 Aug 2014 13:56:11 +0000 (15:56 +0200)]
Setting tag tests-5.3-8
add a few test steps to check that a slice can be emptied using an empty rspec
Thierry Parmentelat [Wed, 20 Aug 2014 11:03:19 +0000 (13:03 +0200)]
test that a slice resources can be released using an empty rspec to allocate and provision, and that this propagates to the plc properly
Thierry Parmentelat [Wed, 16 Jul 2014 21:19:01 +0000 (23:19 +0200)]
Setting tag tests-5.3-7
run linux32 in 32bits VMs
Thierry Parmentelat [Mon, 14 Jul 2014 08:48:04 +0000 (10:48 +0200)]
velvet needs attention
Thierry Parmentelat [Sun, 13 Jul 2014 11:34:54 +0000 (13:34 +0200)]
fix for calling stuff on the guest side with the right personality
Thierry Parmentelat [Wed, 4 Jun 2014 08:20:28 +0000 (10:20 +0200)]
Setting tag tests-5.3-6
split sfa-create-slice into 3 distinct steps sfa-rspec sfa-allocate sfa-provision
Thierry Parmentelat [Tue, 3 Jun 2014 14:52:12 +0000 (16:52 +0200)]
Merge branch 'sfa-geni-v3' of ssh://git.onelab.eu/git/tests into sfa-geni-v3
Thierry Parmentelat [Tue, 3 Jun 2014 14:50:07 +0000 (16:50 +0200)]
split sfa-create-slice into sfa-rspec\ 1 sfa-allocate and sfa-provision
Thierry Parmentelat [Mon, 2 Jun 2014 07:49:36 +0000 (09:49 +0200)]
Setting tag tests-5.3-5
dropped the pl sfa rspec_style in default config
dropped slice manager, sfi talks to AM directly
bump ram to 1.5Gb
Substrate probes for total memory
rename sfa-add-* steps into sfa-register-*
new steps sfa-remove-user sfa-insert-user-in-slice sfi-show-slice sfi-show-slice-researchers
sfi-show-site slice-fs-deleted slice-fs-present super-speed-up-slices
new macros sfa-populate sfa-delete sfa-user-slice slice-up slice-down slice-up-down
sfa-get-expires pending
Thierry Parmentelat [Wed, 28 May 2014 15:39:41 +0000 (17:39 +0200)]
new step sfa_get_expires is able to retrieve the actual expires timeout in a slice
needs to be welded in sfa_renew_slice so we really check renew does what is expected
Thierry Parmentelat [Wed, 28 May 2014 09:07:42 +0000 (11:07 +0200)]
only test the pg rspec_style
Thierry Parmentelat [Wed, 28 May 2014 09:07:22 +0000 (11:07 +0200)]
oops, should be part of the previous commit, sfi needs to be redirected to the port from config
Thierry Parmentelat [Wed, 28 May 2014 08:44:26 +0000 (10:44 +0200)]
considering SM as deprecated :
tests will use sfi connecting directly to the AM, and turn off the SM
Thierry Parmentelat [Tue, 27 May 2014 08:41:46 +0000 (10:41 +0200)]
do not ignore the ssh-slice-again step; this is still broken in f20
Thierry Parmentelat [Fri, 23 May 2014 13:36:12 +0000 (15:36 +0200)]
super-speed-up-slices
Thierry Parmentelat [Thu, 22 May 2014 16:10:39 +0000 (18:10 +0200)]
hide enfoui
Thierry Parmentelat [Thu, 22 May 2014 14:39:22 +0000 (16:39 +0200)]
remove warning that is mor econfusing than helpful
Thierry Parmentelat [Thu, 22 May 2014 11:52:35 +0000 (13:52 +0200)]
enfoui is back in business as a kvm box
Thierry Parmentelat [Thu, 22 May 2014 10:45:42 +0000 (12:45 +0200)]
RAM for test nodes = 1.5 Gb
adjust max kvm nodes per box accordingly
Thierry Parmentelat [Thu, 22 May 2014 10:31:43 +0000 (12:31 +0200)]
probe for boxes memory
Thierry Parmentelat [Wed, 21 May 2014 13:18:16 +0000 (15:18 +0200)]
increase memory for test nodes to 2Gb
Thierry Parmentelat [Wed, 21 May 2014 08:15:34 +0000 (10:15 +0200)]
CompleterTask used to have a failure_message mechanism, then we added failure_epilogue, now we remove failure_message
Thierry Parmentelat [Wed, 21 May 2014 08:05:26 +0000 (10:05 +0200)]
cleanup slice-fs-* steps, nicer output, and shows contents when it's wrong
next step is to unify failure_message and failure_epilogue
Thierry Parmentelat [Tue, 20 May 2014 14:19:25 +0000 (16:19 +0200)]
oops again
Thierry Parmentelat [Tue, 20 May 2014 11:57:27 +0000 (13:57 +0200)]
oops
Thierry Parmentelat [Tue, 20 May 2014 11:05:20 +0000 (13:05 +0200)]
this is how TestMapper deals with subfields
Thierry Parmentelat [Tue, 20 May 2014 09:07:40 +0000 (11:07 +0200)]
isolate settings for plc-config-tty or sfa-config-tty in a separate 'settings' dict to avoid the necessity to repeat these in TestPlc.py
some were missing like PLC_HRN_ROOT and PLC_SHORTNAME
Thierry Parmentelat [Tue, 20 May 2014 05:30:26 +0000 (07:30 +0200)]
fix decorator slice_mapper__tasks so that steps can be ignored