Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
JIRA: DOCTOR-79
Change-Id: I9193ec387af3b593b8f19d8f4b72ddec5c759bc0
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
These functions will ease human readability, writing and debugging of
our test codes. Common functions imported from devstack project.
This is series of patches to refactor the test baseline code in what it
will hopefully become more modular, readable and maintainable.
JIRA: DOCTOR-71
Change-Id: Icbd7a1c2b3979081db8e5de3c46d9827ab54d7ca
Signed-off-by: Carlos Goncalves <carlos.goncalves@neclab.eu>
|
|
As getting servers list under all tenants in specify host takes most
of the time when inspector handling the host failure event, cache the
host-vms list when inspector start run to improve the performance.
add receive event time in inspector
Change-Id: I5fca2816ae9a176be4791bf5dbe26b1c30e743c5
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
|
|
When ran in functest on arm, the doctor suite fails because the x86
image is hardocded in the source files. This change adds the option to
run the doctor suite with an existing image as opposed to downloading
and creating it.
The image name is taken from an enviroment variables (if it is set).
When ran with functest, the env variable is set in doctor.py file.
Change-Id: Ib6ce56e650ebe8b29618a9a31ec990f9f0895f54
Signed-off-by: cristinapauna <cristina.pauna@enea.com>
|
|
Method 'wait_for_status' in OpenStack CLI test_server.py suggests that
we might have to wait a little bit longer after server reaches expected
status. Better wait than break.
Change-Id: I4aaa3091ac2e458aa527b6bd0164d65d46c85017
Signed-off-by: Carlos Goncalves <carlos.goncalves@neclab.eu>
|
|
Change-Id: Ib85571c3939ecb5136eba2c5de8d4b72f2d0659e
Signed-off-by: Carlos Goncalves <carlos.goncalves@neclab.eu>
|
|
Change-Id: Ib7462570b80bd0cf8214411a3467c5484a8d98a1
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
Change-Id: I0d5aac135d66bb41b4ecfe651f071800c323902d
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
Change-Id: I1e3ef0ed1483de6e0b4c4a4101b509a3a3ec2465
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
Change-Id: I59a2b3893272c8cb44d92babb8d2d7b0c0cf3302
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
get installer info and get key to access cloud can be picked
up as common function for prepare env before test
Change-Id: I1a96c4ad12f3f36dec0291c2f458e97e85f3ca27
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
When running Congress as Inspector implementation, the Monitor has to be
started after starting the Inspector because we need to first ensure the
Doctor datasource is created, otherwise the Monitor cannot get the
Doctor datasource ID at init.
This patch defaults the Inspector to 'sample' and for the time being
functest will run only against 'sample', not all supported Inspector
types ('sample' and 'congress'). Testing multiple Inspectors in single
functest run would require major additional changes to our test scripts.
It should still be done and addressed in a future patch. This patch
focus on adding testing support against Congress as first step.
One can test against Congress executing for example:
$ INSPECTOR_TYPE=congress INSTALLER_TYPE=local COMPUTE_HOST=compute1 ./run.sh
JIRA: DOCTOR-56
Change-Id: Icebd6fd6ad0c01d511c97e804727ad2a71f742e8
Signed-off-by: Carlos Goncalves <carlos.goncalves@neclab.eu>
|
|
Change-Id: I9a85923b1b1766435280aeddfcebfaccbbb4e098
Signed-off-by: Ryota MIBU <r-mibu@cq.jp.nec.com>
|
|
Change-Id: I5c449ac13d804aa1a65c703fc1e6b0caa70ff142
Signed-off-by: Ryota MIBU <r-mibu@cq.jp.nec.com>
|
|
Now ssh tunnels will be created from each controller node to
functest container, so that aodh can send message to doctor
consumer in functest VM.
Change-Id: Ia4360e78caa99b17a27802477765ec055c1144d7
Signed-off-by: Ryota MIBU <r-mibu@cq.jp.nec.com>
|
|
|
|
this patch removes unnecessary single quotes
JIRA: DOCTOR-63
Change-Id: Ie24cb4c3c8fcae43371e58068cf03ee74ce9962e
Signed-off-by: Ryota MIBU <r-mibu@cq.jp.nec.com>
|
|
|
|
The doctor test kick by functest fails in some deployment due to no
connectivity from controller to the doctor consumer.
To avoid this issue, this patch add ssh tunnel from installer node to
the functest contatiner.
JIRA: DOCTOR-63
Change-Id: Ide85137767382a53ebd15b875fd216aeed632705
Signed-off-by: Ryota MIBU <r-mibu@cq.jp.nec.com>
|
|
sometimes after create `admin` role, we use `openstack user role list doctor
--project doctor` cli can't get the `admin` role. We should add project when
we create user.
Change-Id: Ib3ee8a3009600aa2c88de2350b483f295c12d8d5
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
misuse `ssh_options` which is not define in script
Change-Id: I932859750d7fbe7cd4b69e3f3aceb5d84fa8f485
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
Now, we can see log messages in INFO level, but it's not clear from
where to where comes from which log file.
This patch adding file names in INFO level.
Change-Id: I3c8fa9f28c2c9ca3e758536243a8110849591825
Signed-off-by: Ryota MIBU <r-mibu@cq.jp.nec.com>
|
|
This patch extends support for fuel OPNFV installer for the test
scenario.
JIRA: DOCTOR-62
Change-Id: I09746b0e64b3999363db53c318027b81c471e2c4
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
JIRA: DOCTOR-55
Change-Id: I41dc85fe43c2c8f9c5d98ff33770a59483caa19a
Signed-off-by: Carlos Goncalves <carlos.goncalves@neclab.eu>
|
|
Recently functest container had its python requests lib updated to version
2.8.0 and it no longer supports passing in header values of float type.
Change-Id: I31da351d7f5c7c1c3ff5121b7b000f3bab1d76bf
Signed-off-by: Carlos Goncalves <carlos.goncalves@neclab.eu>
|
|
JIRA: DOCTOR-55
Change-Id: I36a8a3eaeceb3a6baacad44f410b8464ba7704d5
Signed-off-by: Carlos Goncalves <carlos.goncalves@neclab.eu>
|
|
- support CI_DEBUG=true and CI_DEBUG=True
- use fqdn for hostname matching in overcloud, but keep using short
hostname for undercloud
- add printing of paramters, so that we can see them in info level
- make a result failed if the consumer didn't get message
- allow host_state=UNKNOWN after failure injected
JIRA: DOCTOR-63
Change-Id: Ib9bf02df3a68bec03e53db9d51c201545cf5a912
Signed-off-by: Ryota MIBU <r-mibu@cq.jp.nec.com>
|
|
|
|
use `ip route get $COMPUTE_IP` to get CONSUMER_IP. And add script
fail if it failed to get CONSUMER_IP.
Change-Id: I2e330c9be6e00f31e6c3e27dd674f9aabbb74a41
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
JIRA: DOCTOR-63
Change-Id: I4bc932dac22aca746489910bccb87f6571e3200b
Signed-off-by: Carlos Goncalves <carlos.goncalves@neclab.eu>
|
|
Extending sleep time by 60 seconds to allow Nova server to sync-up
states with nova-compute service.
JIRA: DOCTOR-66
Change-Id: I71c2f186fb3faf5550c7aede465c93c0674943e2
Signed-off-by: Carlos Goncalves <carlos.goncalves@neclab.eu>
|
|
|
|
'openstack server delete' tries to delete VM with admin credentials
which fails because $VM_NAME is under a different tenant.
Same for ceilometer alarms that we created with doctor creds.
Change-Id: Ibba3b842dc8fa160137bb04316d285e52d776641
Signed-off-by: Carlos Goncalves <carlos.goncalves@neclab.eu>
|
|
We need to add the `debug=True` in consumer to get the debug log
to get the time of consumer received the notifier
Change-Id: I90f2f8167a8670e28a3a5bcb5af054d378ef9b5b
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
- revert to use nova command to get IP address of compute host
- disable interface which has the IP address of target compute host
JIRA: DOCTOR-63
Change-Id: Ic5dc7706476a542f13a942fdae5b347ce28d3f72
Signed-off-by: Ryota MIBU <r-mibu@cq.jp.nec.com>
|
|
|
|
|
|
|
|
Parameters defined in another shell cannot be used in the main shell.
This leads bug in funcsion which execute another shell to performe
openstack command as doctor test user.
This patch makes the test script use --os-* options to performe as
doctor test user, so that we don't have to execute new shells in
the test script.
Change-Id: If148f00acae919a926bb5e9f2af8dd433a4110f5
Signed-off-by: Ryota MIBU <r-mibu@cq.jp.nec.com>
|
|
Each installer has different compute users and we set a default for each
in get_compute_host_info. Having COMPUTE_USER set at the beginning as is
and not passing in a value will default to 'none' and thus overriding
the default user of the installer. For example, COMPUTE_USER would be
set to 'none' instead of 'heat-admin' for Apex installer.
Change-Id: Ia60a88f68c52604bac6b7eef18e97e11328f1a6e
Signed-off-by: Carlos Goncalves <carlos.goncalves@neclab.eu>
|
|
This patch is to make sure the test will be stop and report it failed.
JIRA: DOCTOR-63
Change-Id: Ibab577a100dee784df14e8b129fbde626d04fce9
Signed-off-by: Ryota MIBU <r-mibu@cq.jp.nec.com>
|
|
we can not use `localhost` in `--alarm-action`, it should be the IP
which the consumer started on.
Change-Id: I50b99e4ff7609c474fb569049b3b8cd216c9178b
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
Change-Id: I335cf82f3996359327d0cacfcdc53eda8a709f86
Signed-off-by: Carlos Goncalves <carlos.goncalves@neclab.eu>
|
|
Change-Id: I87c971fedc6106165bfe0716a475edad5c43640c
Signed-off-by: Carlos Goncalves <carlos.goncalves@neclab.eu>
|
|
In `cleanup` function, we can't delete doctor project and doctor user
before checking host status. Because we use API to get the `host status`
under the doctor project.
Change-Id: I38c0929eeff9378051e0dcda308734ab7b353509
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
|
|
If there are multi-compute node, vm may not be scheduled to the
specified compute host. So get the compute host after vm is started.
Change-Id: I1bc3caaf8305beded9eb297b116968a190fd9751
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|
|
We should add the version para to use `openstack server show` cli
to get `host_status`
Change-Id: Ie0332e3766cb88f687fa66bcd5d8218f311e8add
Signed-off-by: dongwenjuan <dong.wenjuan@zte.com.cn>
|