| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
| |
Change-Id: I4a2ac7b819929c2de6c8ea5520257800d2e134cc
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
|
|
|
|
| |
Create network side failure by opening and closing the ports
related to gluster-blockd during creation of the PVC's,
Verify PVC's are bound and validate multipath.
Change-Id: Ibc53a13e2abb8674661da83d5881a13bbb2ad7fb
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In OCS 3.11.4 was changed the approach for 'provisioners' in storage
classes due to the BugZilla bugs [1] and [2].
So, change the approach for getting 'provisioner' value in this project
making it be autocalculated based on the data taken from the
block provisioner pod, which is located in the working namespace.
[1] https://bugzilla.redhat.com/show_bug.cgi?id=1732787
[2] https://bugzilla.redhat.com/show_bug.cgi?id=1738394
Change-Id: Ic1037038e39abba2b2a1ad6dc3ddbfe42473cb88
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
|
|
|
|
|
| |
Define 'check_heketi_db_inconsistencies' config option setting
there 'False' or 'True' values. Default value is 'True'.
If heketi client doesn't support the 'heketi db check' feature, then
'heketi db check' just won't be performed.
Change-Id: I7faff35b15e40d864c0377ae7fee154e217d8eae
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This change required due to -
1. Get block volume by prefix is common step which is used
in two places for now and will be used in other places
too.
2. Hence add library "heketi_blockvolume_list_by_name_prefix"
in heketi_ops.py.
3. And use added library and update code from class
"GlusterStabilityTestSetup" and "TestGlusterBlockStability".
Change-Id: I9e60d58d5c84380104081335745270b3d21ff071
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
|
| |
Change-Id: I22410ad8c81ca71ac9c2cbaddd66f086abcdf3c5
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
|
| |
Change-Id: I1055b232711c95fdbf89ba9fe8b9d61a561fbcd5
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
| |
Change-Id: I96759bbcd74cad7755c99255b4b96261133d5adc
|
|
|
|
|
|
|
|
|
| |
As of now user can only pass dc name as argument
to validate iscsi and multipath session, with this
support user can also pass pod name by specifying
rtype=pod
Change-Id: Iecc86edc2a2239fc05535f0e4f7e9ae93d5e76f5
|
|
|
|
|
|
|
|
|
|
| |
Create 8 Block PVC and wait for them to get bound. Delete created
PVC's and create 8 more PVC's simultaneously. Kill the tcmu-runner
service while PVC creation and deletion is in the progress. Restart
all the services related to block volumes and verify all pending
operations are completed sucessfully.
Change-Id: I0cb4cd29b92233a65be93f4b96f1a9a0cb8bed9f
|
|
|
|
|
|
|
|
| |
Add library to check db results
Validate count of volumes of heketi and db are same
Change-Id: I6e75f2d8d5f19a94c1dced5a07f9c948b4de9ef4
Signed-off-by: Sri Vignesh <sselvan@redhat.com>
|
|
|
|
| |
Change-Id: I46ad6aefd3c699b4ce7807db89c2016791a48e05
|
|
|
|
|
|
|
|
| |
Add new function called 'wait_for_pvcs_be_bound', which allows us
to wait for 'Bound' state of multiple PVCs using single timeout.
Also, reuse this new function in the baseclass.py
Change-Id: Ia3809d322925b90fdef984db59b3e9e83a82abc7
|
|
|
|
|
|
| |
Change the 'cluster' to 'clusters' in kwargs description.
Change-Id: Iac9170d0549adc6bb5b76f5772de47d1db083951
|
|
|
|
|
|
|
| |
Remove json description from heketi_device_add library as it is not
supported.
Change-Id: If3121294dc9456587af36d65726842f399730a9b
|
|
|
|
|
|
| |
This reverts commit 046244942f7ad3f5f24b675dfa299ad955f1412f.
Change-Id: I1f49d613af3dd236bf4113db757cbf9ce752e973
|
|
|
|
|
|
|
|
|
|
| |
Add new function called 'wait_for_resources_absence',
which allows us to wait for absense of multiple resources
using single timeout.
Also, reuse this new function in the baseclass.py for PVC/PV
removal waitings.
Change-Id: I6e3daae012199e50d3aeb68b697cfae801b0b49b
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fix consists of -
- Use "**kwargs" approach in the "heketi_node_list"
function as it is done in lots of other functions.
- Parse the CLI output in the "heketi_node_list" function
using regex instead of the splitting and stripping strings.
- Combine test cases related to the same feature into one
module - test_heketi_node_operations.py
- Remove redundant checks which already exist in common
libraries.
- Remove unnecessary logging.
Change-Id: I815ddfbbacb765140229e7630ec87a6bbaa6255b
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
| |
Change-Id: I4a6fd02f5713b4e86a9d70cdaffd3ca9369932f4
|
|
|
|
|
|
|
| |
Add new TC verify PVC's create and delete fails while gluster-blockd
and tcmu-runner services are down on one node.
Change-Id: I5acbcfd6c9a227c0cce21c62f5e162caf970aa22
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fix consists of 2 parts:
- Calculate correctly left time for waiting in
'scale_dcs_pod_amount_and_wait' function. Before, it had bug,
where we were waiting less time than requested. And it led to the
unexpected timeout errors.
Example: it could fail waiting only 180sec having 600sec as real
timeout.
- Reset 'attempts' on the instantiated 'waiter's to avoid
redundant waiting reusing the single waiter instance.
Example. On deletion of 20 PVCs we save about 1,5 minutes.
In whole test suite we create much more PVCs than 20.
Change-Id: I5d06a63dd0c2c5bd67fdb09fef87948d65e6bf22
|
|
|
|
|
|
| |
As part of the fix loaded the output into json which was missed.
Change-Id: Ia3d07f768362232ec2b34641be1a6ae4c4eec399
|
|
|
|
|
|
|
|
|
|
| |
Add 10 test cases for testing Heketi zones feature.
To make all the Heketi zones test cases run in the single suite using the same
cluster set 'common.allow_heketi_zones_update' config option to "True" value.
By default it is set to "False" and only small set of test cases can run
having such value.
Change-Id: I69a1f7c96c9f52a06134e715e113ccd9b06764e6
|
|
|
|
|
|
|
|
|
|
|
| |
In test case when we use large disk size, heketi takes time to create
volume and due to timeout it gives error.
Add fix in such test cases to check if volume created after getting
an exception, get details of such volumes or raise exception in case
it fails to create volume.
Change-Id: I1c23a8c6558c23edf8947771e4f41a4bd3ffd66a
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
|
|
|
|
|
| |
Gluster block test cases failed with error "self.node" variable not found
as self.node variable is not created in "GlusterStabilityTestSetup" class.
Fixed change in baseclass.py by using variable "self.ocp_client[0]".
Change-Id: I2affabbb626b7e266a9bb243d3ec249229ec9670
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
"get_ocp_gluster_pod_details" library gives error
"IndexError: list index out of range" in case of
independent mode setup as "get_custom_resource"
returns "[[]]" which in not none.
Added fix to check if any element present in list.
Change-Id: Iffd081f835ffaf9cb50c020cd297444ea2678950
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
| |
Change-Id: If30737958bc667264fc01fe81d411f406b501918
|
|
|
|
|
| |
Change-Id: I5886680a0d5666c68c677893e0fb327be0e80760
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
|
|
|
|
| |
Move get_block_hosting_volume_by_pvc_name from
test_restart_gluster_services to baseclass to
be able to reuse it in other test modules.
Change-Id: I65847792601b422293d4baed2ccade664ad7e54b
|
|
|
|
|
|
|
| |
Add new TC where we verify rediscovery of iscsi paths
which happen after abrupt reboot of initiator node.
Change-Id: I841e875881c47f8215d48821cd76c0399d43badc
|
|
|
|
|
|
|
| |
Added Error msg in GlusterBlockBaseClass for better visibility in case
of failures or debugging purpose.
Change-Id: Ia92cd6f6129cb5aa55f6f8e807ef056e54691956
|
|
|
|
|
|
| |
Changed the description of libraries in gluster_ops where it was wrong.
Change-Id: I8ea9ec4ff1f28250d4544d6329b386271cf1e551
|
|
|
|
|
|
|
|
| |
modification in block volume name parameter passed twice while
creating
Change-Id: Ieb827a8e7fa40a84eed7f8c5e90760710e6615b2
Signed-off-by: Sri Vignesh <sselvan@redhat.com>
|
|
|
|
|
|
|
|
| |
We have test cases which create more than 1 app DC at once.
So, add functionality to be able to create bunch of DCs in parallel
and reuse it in one of test cases.
Change-Id: Id606d02c31a919bbc6d49d59714dd5628c6a835d
|
|
|
|
| |
Change-Id: Iaa7deca275958d4de68601dc16d1920f3dab85f2
|
|
|
|
|
| |
Change-Id: Ia0325c4309837fe3f89eab3c066775f9fb1ab1de
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
|
|
|
|
| |
Fix error handling logic in the "vmware" cloud provider
module by adding "raise" statement in places where
it is lost.
Change-Id: I92e97da4109bc6ab7368b41d2a6886e9f2be31c1
|
|
|
|
| |
Change-Id: I2ebc1c2e7e7aefb5e0d70342a7ec243a12b0663b
|
|
|
|
|
|
|
|
| |
Add fix to test cases which are skipped due to bugs -
BZ-1644685, BZ-1652913, BZ-1634745, BZ-1635736, BZ-1636477, BZ-1641668
Change-Id: I03a7f7dddaba5d3e41770969352b2263c84cb920
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
| |
Change-Id: I4196cb1395a3720e03f62473d11e5f46d797c355
|
|
|
|
|
|
|
|
|
|
|
| |
New TC insures that app pod is restarted properly when one of the
target node is down.
This patch includes libs of vmware api
These libraries can perform VM operations like power on and off via
vmware client api's.
Change-Id: I11ad4dc3318cb0583de5882d8067ed7e30ea9962
|
|
|
|
|
| |
Change-Id: Ia4e0fb737b16ea7bdc8ffd5ae44cdd418471552a
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
|
|
|
|
|
| |
Separate node reboot functionality from single test case to re-use
it at other places. Update test case accordingly.
Change-Id: Ib9a7f15d29237e4f21aafc408c074e799e706740
Signed-off-by: Sri Vignesh <sselvan@redhat.com>
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
|
|
|
|
|
|
| |
"get_gluster_pod_names_by_pvc_name" function depends
on the ocp3.11 specific shell command, which fails on
old OCP versions. So, fix it by getting info in compatible way.
Also, update usages of this function with new return data structure.
Change-Id: Ibb8559590a1288c032630b3292f631d28bc87263
Signed-off-by: vamahaja <vamahaja@redhat.com>
|
|
|
|
|
| |
Change-Id: I7ead23c46a472fee70d684c45f32f5e4efb0674f
Signed-off-by: kasturiNarra <knarra@redhat.com>
|
|
|
|
|
|
|
|
| |
This test case verifies that when tcmu-runner is down we are able to
run IOs on block volumes, and stoping tcmu-runner would effect
gluster-blockd and gluster-block-target services.
Change-Id: I0b4a23f7c2dce909f07a22893f83a6c1d0285091
|
|
|
|
|
|
|
|
|
|
| |
Before, in case we had some constant networking issue trying to connect
to a host, we were falling into the endless loop of connection recreation.
The error we got after it is the 'recursion limit exceeded'. It is unexpected.
So, change this logic making it avoid recursion and do just one
connection recreation attempt which is useful to fix broken connections.
Change-Id: Id808edbad7e6d69ad58a75bfbae176fddb173d18
|
|
|
|
|
|
|
|
| |
It happens that heketi client located out of the Heketi POD may fail
not reaching the server side. So, add back-up approach where we run
Heketi commands on a Heketi POD when main commands fail.
Change-Id: Ie6ae5be82082f34426f9288b02575e3abd4940f5
|
|
|
|
|
|
|
| |
New TC verifies that it is possible to run IOs on block
volumes when gluster-blockd service is down.
Change-Id: Ia1de14d2990f833221f6725e9b0e48d77ef85c10
|
|
|
|
|
|
|
| |
move gluster-block stability TC to new module named
'gluster_stability/test_gluster_block_stability.py'.
Change-Id: Iac1f432e438a2815bdc2115ab19e0170e46930c1
|