* modprobe: only reload interfaces for network drivers
hv_netvsc needs a special case, since reloading it has the potential
to leave the node without a network connection if things go wrong.
For most modules this isn't neccesary, so just special case netvsc.
* kata conformance automation
* Flak8 formatting changes
* isort fix
* changes for CH perf metric test
* change for making azure package optional in poetry.lock and removing unneccessary lisa tools from kata conf test
* formating wth make fix
* PR review ==> remove CH perf metric changes, Go tool version changes, Path changes in shell.py, log changes in testsuite, made azure packge optional
* CH Perf metrics testcase integration
* remove kata related changeg to separate the CH perf PR
* remove git related changes to separate the CH perf PR
* PR review changes : Added Error in subtest, created different function to run perf-metrics test
* flak8 and isort formating
* PR Review change suggested by Anirudh, removal of unwanted logging, adding regex string for referral, removed unwanted comment
* PR Review suggestions by Anirudh : Dataclass implementation, report file to logpath
* removal of unwanted log
* lofa path change fore report file
* PR Review by Anirudh : nit changes
* PR Review by Chi: Pushed notification logic to single loop in _create_perf_metric_report function
* flak8 isort fix
* Change for quick send of test msg result and removal of dumping result to json file
* Changes to remove unused import,white spaces and BLK100
* flak8-isort change : add arg's datatype for function
* PR Review by Chi : return code handling change, log level change in case of exception to info
* removal of dataclass CHPerfMetricTestResult
* Changes for regex to get metrics
* Removal of result dict and using variable
* BLK100 Black format changes
* addition of None check in regex search
libvirtd logs are useful for debugging libvirt related issues that might
happen while executing testcases. So, retreive the logs using journalctl
and write them to the host node's log path. Also introduce a new tool
for working with the `journalctl` command.
Signed-off-by: Anirudh Rayabharam <anrayabh@microsoft.com>
* DPDK: add close() call before netvsc reset in rescind kill
node.close() seems to be more reliable for killing the unresponsive
process, so attempt node.close before resetting the network interface.
after_case already performs this reset anyway so doing it twice it
pointless if we can avoid it.
* clean up comments
Two small fixes:
First and last samples have bad data, since we're looking for an order
of magnitude difference they can mess up the data sometimes. Discard
first and last sample before calculating the mean of the data.
Fix an incorrect assumption about the boolean value of an empty
iterator. Cast to list first, makes other logic cleaner.
The awaitable timeout happens on different places, it's not easy and
unnecessary to check the timeout state together. So use a collection to
track them independently.
Introduce a new test suite for testing the MSHV root partition. Add a
test case that creates and destroys multiple VMs in parallel and repeats
this multiple times to stress the virtualization stack.
Signed-off-by: Anirudh Rayabharam <anrayabh@microsoft.com>
Lili Deng
mcgov
Smit Gardhariya
Anirudh Rayabharam
Sonia Sharma
Divyansh
Purna Pavan Chandra
Chi Song
Divyansh Srivastava
lubaihua33
Kameron Carr