| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
| |
User's home dir in CRS logic was hardcoded as "root".
So, fix it by calculating home dir of the current user.
Change-Id: Ib586fa605390126eb3a2efaf8c93856e52676aaf
|
|
|
|
|
|
|
|
|
|
|
| |
Renamed cns_baseclass.py to baseclass.py with updated BaseClass.
Merged HeketiBaseClass to BaseClass with appropriate changes.
Updated delete_volumes and delete_block_volumes methods used in
test case method with heketi_delete_volume and heketi_blockvolume-
delete.
Change-Id: I8f623f1692cd863dc8ff041f7e8525ee0dc84f7b
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Before, 'heketi_version.py' module didn't consider version of the
'heketi server' package on the Heketi POD. But it is more valueable than
depend on the client only. So, make all the requested version
comparisons be applied to the server's version.
Interaction with the module's logic stays the same, just one behavioral
thing changes: now, if client version is older than server's one, we
raise exception.
Change-Id: I317102556d4a5f51607c3b9da93da0d4a3b74f7f
|
|
|
|
|
|
| |
To let system services start up in time.
Change-Id: I4b4781119656c9fdc75017a28c80c20bdd0eae21
|
|
|
|
| |
Change-Id: I5f25b1dfce5d3298295939f59c3ff435ae7c6d8a
|
|
|
|
|
|
|
|
| |
For the moment, if subscription manager fails, we continue
deployment, but it is mistake. We just spend redundant time
before we try to install anything.
Change-Id: Ib95c00a3c605b98f3cd6b7998d4b482533a5d491
|
|
|
|
| |
Change-Id: I51056965a55f73b0cb4b7dac8f1d4234f2f66528
|
|
|
|
|
|
|
|
|
|
| |
Before, we were installing latest heketi-client version on master nodes.
But, in case we install old OCS/CNS we may end up with false
impression that we have new features, when it is not so.
So, read heketi-client package version from Heketi POD and install
exactly the same version on the master nodes.
Change-Id: Ic4e9edd7f6ef614d28c061712473aaf9b9f652e2
|
|
|
|
| |
Change-Id: I40b7ea79d1f7dbc82db825f6ced4447a157361e6
|
|
|
|
| |
Change-Id: I91f916e31bb23197dd972a17dd3a5ee62ee04b77
|
|
|
|
|
|
| |
These two files were missed due to improper .gitignore file data.
Change-Id: I49f396c1c1d37fcbd4257bf35aaee75f41b099bc
|
|
|
|
| |
Change-Id: I67a6306fa6cd2b5c5c886efbdbdd5a7134b3225e
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add end-to-end deployment tool of OpenShift and OpenShift Container
Storage on top of VMWare.
Added code is modified version of the
'reference-architecture/vmware-ansible' dir from the following repo:
https://github.com/vponomaryov/openshift-ansible-contrib
Read 'deployment/README.rst' file for more details about the
deployment tool.
Change-Id: Ic96f252ff786cc1ecf24d27f0ec47e324131e41b
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In some cases, it is useful to stop test execution after first failure.
Not only stop it, but also skip all the scheduled cleanups.
It will allow us to keep a cluster in the best state for a root cause
debugging.
Just define 'common.stop_on_first_failure' option to True value
in the config file. It will make the test runner stop after first
failure skipping all the scheduled cleanups and pending test cases.
Change-Id: I963eb038a11a8e2088a84f7ba4838870ea3e657a
|
|
|
|
|
| |
Change-Id: I7f4fea8f197294f2df8bba19c8c0f0422b62892a
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
|
|
|
|
|
| |
Add Heketi version checks to the test cases and Heketi libraries.
In case of incompatible Heketi version in a setup,
test cases will be skipped and direct usage of Heketi libraries
will cause them to raise the NotImplementedError exceptions.
Change-Id: I462e76f615165140f2a8c7e796c1c582b8f4dc91
|
|
|
|
|
|
|
|
| |
This library is helpfull for diagnosing output of gluster and heketi
and it compare both the outputs and the test case verify this feature
is working properly.
Change-Id: Ice607003f32db2bce2f540a7651024b66b071634
|
|
|
|
|
|
| |
To be able to perform safe Heketi client package version comparisons.
Change-Id: I64f0a55f988fc1446b133556c395f0640d65ada1
|
|
|
|
|
|
|
|
| |
In case of network problems, "curl" command may stuck in
a retry-loop. So, add explicit time-limit as 10 seconds to avoid
endless waiting of the functions which use it.
Change-Id: Ia2d94358e8f978cb73360535167c703da2636874
|
|
|
|
|
| |
Change-Id: I9ab3c6bea56a02f76ce818ee095ffe4afc0e64c0
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
|
|
|
|
| |
Trying to create pvc with defined arbiter file size and
verifying that when we are expanding volume it is expanding
according to avg file size
Change-Id: I5c39339f70702bf685c237ae16e21f54384e51db
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This test case talks to the Gluster PODs directly. It is not
compatible with Standalone Gluster clusters.
So, fix this test by reusing special common function,
which allows us to abstract from Gluster cluster deployment type.
Change-Id: Ib131984ac0a4623d1d847f9961be26da573d8513
|
|\ \ |
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This test case expects that Gluster cluster has the same amount of
volumes as Heketi does. But, it is wrong expectation.
We may either create gluster volumes without Heketi or have
mismatches after some other Heketi failures.
So, fix this test case by comparing starting and resulting amount of
volumes in Gluster and Heketi.
Change-Id: I322c922fbdf102ccc5fb9f2145eef7b3f1d1a80a
|
|\ \ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
For the moment, we have 3 different test cases which use
'additional_devices' config option. All of them will fail in case
of specific structure of config.
So, make those test cases be smart enough to understand absence of
such option data and skip test cases with useful info about it.
Change-Id: Id31de7e20188041d2148edf37b3f8c95645c5e52
|
|\ \ \
| |_|/
|/| | |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Removing this code as it is in additional library file, requires to
pass config through config file and requires to upload the scripts.
Also it contains hardcoded credentials.
Will add new function in openshift_ops.py library similar to
"oc_create_app_dc_with_io" with new updated template.
Change-Id: I7cfce3fe23b222d40bff1c829afa97f2ffcb621d
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|/ /
| |
| |
| |
| |
| |
| |
| | |
"heketi-cli" requires quotes for "--gluster-volume-options" arg.
Without it command will fail. So, add missing quotes to make it work.
Change-Id: I17fddfe653dceaa3c6b56f2a473ca7798de02698
Signed-off-by: Sri Vignesh <sselvan@redhat.com>
|
|/
|
|
|
|
|
|
| |
In this tc we filter pod by dc name with hard-coded project name.
It is completely incorrect approach.
So, fix it by getting project name from config.
Change-Id: I062bc22c9cbf22bbb101a48bad94d0540a3c4bb8
|
|\ |
|
| |
| |
| |
| |
| | |
Change-Id: I2379043b771cc654eb227d457223b3a24434fd21
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|\ \ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
For the moment, we write events of failed PVC only to glusto logs.
And only info of one type of events - 'ProvisioningFailed'.
But in case of CI, it is not really suitable. We need to have info in
the exception message.
So, add all the pvc-related events to the message of raised exception when
PVC fails to reach the "Bound" state.
Change-Id: I60e3b9a86acdf251d282059cd2e494c70ae07c70
|
|\ \ \ |
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
This test case has fragile logic. It doesn't assume
we expect new block hosting volumes to be created when we calculate
the size for block volume.
So, fix it by doing 2 things:
- Use total block hosting volume size, not it's 'free space'.
- Get info of block hosting vol if block vol gets created unexpectedly,
and read it's size. If the size is bigger than block vol's size, then
pass the test.
Here, we assume that existing, during calculation,
block hosting volumes were not the biggest possible and
autocreation is enabled in addition to the presence of
the available size on the Heketi devices.
Change-Id: Ic7d5325b1144dce0e56999617f3c6d7c0829e42d
|
|\ \ \ \ |
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Because it was very fragile and skipped due to bug.
So, after bug was fixed we need to enable it back.
But, also, considering it's stability, it required complete change of
logic. So, it is rewritten completely and moved to more appropriate
module where we have other test cases which perform device operations.
Change-Id: I1b8c18c8ea883f83861e81e97c534459d7588175
|
|\| | | | |
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
It was skipped some time ago due to the bug. Now this bug is fixed.
So, we need to unskip it. So, do it, in addition to the required
refactor.
Change-Id: I7c426e9cd3dc38ed48e5b91870bc10bd63998694
|
|\| | | |
| |_|_|/
|/| | | |
|
| | |/
| |/|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Current 'test_add_device_heketi_cli' [CNS-877] tc completely duplicates
another existing tc - 'test_device_remove_operation' [CNS-766].
Being, on it's turn, very fragile and requiring refactor.
So, remove this test as it makes no sense neither to keep it skipped nor
to refactor it.
Change-Id: I40cf7b3d015d1345f0ac2a89569dd64cd67711d3
|
|\ \ \
| |/ /
|/| | |
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
In some rare cases we get following error trying to create distributed
volume:
- Insufficient suitable allocatable extents for
logical volume tp_foo: 25344 more required.
So, fix such situation by catching exception with such message and
retrying with smaller volume size.
Change-Id: I015a4547347858cd8aff9e566e3dc662a94b6b83
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Doing following:
- Rename top-level config option group called 'cns' to the 'openshift'.
- Rename 'cns.setup.cns_project_name' option to the
'openshift.storage_project_name'.
- Rename 'scale' option to the 'openshift.scale'.
- Delete ''devices' and 'zone' options from 'gluster_servers' config
group.
- Delete 'cns.setup.cnd_username' option as unneeded.
- Delete 'cns.setup.cnd_password' option as unneeded.
- Delete 'oc_login' common function as unneeded after deletion of
'cns.setup.cnd_username' and 'cns.setup.cnd_password' options.
- Remove 'Cns' word from base classes.
- Keep backwards compatibility for old options to make transition be
smooth.
Change-Id: I04ddbbad5f64ffeac79a6744480d33a263d63001
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Now, it is possible to run glustolibs commands
on both deployment types: containerized and standalone.
It is possible using 'auto_get_gluster_endpoint' as target for
lib function in addition to the '@podcmd.GlustoPod()' decorator.
Example:
"""
from glustolibs.gluster.volume_ops import get_volume_list
from cnslibs.common import heketi_libs
from cnslibs.common import podcmd
class TestExample(heketi_libs.HeketiBaseClass):
@podcmd.GlustoPod()
def test_get_vol_list_either_on_node_or_on_pod(self):
volumes = get_volume_list('auto_get_gluster_endpoint')
"""
Also, delete all the current usages of 'deployment_type' config option,
and delete the option as unneeded anymore.
Change-Id: I281f287c432a5a9efefda588be436ee285188697
|
|\ |
|
| |
| |
| |
| |
| |
| |
| | |
This test is broken in several places.
So, fix it and also remove redundant steps.
Change-Id: I7a9f2c077dda35a8a477fa27882cddf79ec69ac5
|
|\ \ |
|