On 31/08/12 22:21, Chris Evich wrote:
On 08/28/2012 11:39 AM, Yu Mingfei wrote:
Signed-off-by: Yu Mingfei<[email protected]>
---
client/tests/libvirt/tests/virsh_setmem.py | 98
++++++++++++++++++++++++++++
1 files changed, 98 insertions(+), 0 deletions(-)
create mode 100644 client/tests/libvirt/tests/virsh_setmem.py
diff --git a/client/tests/libvirt/tests/virsh_setmem.py
b/client/tests/libvirt/tests/virsh_setmem.py
new file mode 100644
index 0000000..a9b5155
--- /dev/null
+++ b/client/tests/libvirt/tests/virsh_setmem.py
@@ -0,0 +1,98 @@
+import re, logging, commands, time
+from autotest.client.shared import utils, error
+from autotest.client.virt import libvirt_vm
+
+
+def run_virsh_setmem(test, params, env):
+ """
+ Test command: virsh setmem.
+
+ 1) Prepare vm environment.
+ 2) Prepare libvirt state.
+ 3) Run test command and wait for current memory's stable.
+ 4) Check result.
+ TODO: support new libvirt with more options.
+ """
Good, simple, well written docstring. Thanks!
+ def get_mem(vm):
+ output = libvirt_vm.virsh_dominfo(vm)
+ lines = output.split("\n")
+ get_memory = 0
+ for line in lines:
+ if re.search("Used memory:", line):
+ get_memory = int(re.split(" ", line)[-2])
+ return get_memory
Hmmm, what about making this as a method on vm class? This seems like
it would be useful function? If not, no worries, just a thought I had.
Ok, I will consider about it.:-)
+
+
+ vm_name = params.get("main_vm", "vm1")
+ vm = env.get_vm(params["main_vm"])
+ domid_result = libvirt_vm.virsh_domid(vm_name)
+ domid = domid_result.strip()
+ domuuid_result = libvirt_vm.virsh_uuid(vm_name)
+ domuuid = domuuid_result.strip()
+
+ if params.get("start_vm") != "no":
+ vm.wait_for_login()
Good catch!
+
+ origin_mem = get_mem(vm_name)
+ logging.info("\n======\norigin memory : %i\n======", origin_mem)
I have been in bad mood about too much logging.info() lately. However
I think this one is okay.
+ #Prepare libvirtd service
+ libvirt = params.get("libvirt", "on")
+ if libvirt == "off":
+ libvirt_vm.service_libvirtd_control("stop")
+
+ #Run test case
+ vm_ref = params.get("vm_ref", "")
+ setmem_vm = params.get("setmem_vm", "%s")
+ options_ref = params.get("options_ref", "")
+ options_prefix = params.get("options_prefix", "")
+ options_suffix = params.get("options_suffix", "")
+
+ if vm_ref == "domid":
+ if domid == "-":
+ vm_ref = domid
+ else:
+ vm_ref = int(domid)
+ elif vm_ref == "domname":
+ vm_ref = vm_name
+ elif vm_ref == "domuuid":
+ vm_ref = domuuid
+
+ if options_ref == "default_half":
+ options_ref = origin_mem/2
+
+ setmem_vm = setmem_vm % vm_ref
+ new_memory = options_ref
+ options = options_prefix + " " + str(options_ref) + " " +
options_suffix
I'm nervous about having lots of code between changing libvirtd
service state because if there is any problem, it can mess up all
following tests. I had a big struggle with this while working on new
virsh module.
Can we move all the options/parameter processing code above the
'#Prepare libvirtd service'? I think it will be a little safer this way.
Thanks very much for your remider.
I run all ok in my test, so I do not notice this problem.:-)
+ result = libvirt_vm.virsh_setmem(setmem_vm, options,
+ ignore_status=True,
print_info=True)
Also, '#Run test case' comment probably needs to move closer to here.
+ status = result.exit_status
+ #After set memory, guest's current memory will be set step by step
+ #wait 10s here to be stable
+ time.sleep(10)
+
+ #Recover libvirtd service
+ if libvirt == "off":
+ libvirt_vm.service_libvirtd_control("start")
+
+ if not status:
+ current_memory = int(get_mem(vm_name))
+ logging.info("\n======\ncurrent memory : %i\n======",
current_memory)
Is this too much logging.info()? I can't tell anymore, it's up to you :D
But I just want to print memory before set and after set.:-(
Otherwise, we can not trace memory's change.
+
+ #check status_error
+ vm_status_error = params.get("vm_status_error", "no")
+ addition_status_error = params.get("addition_status_error", "no")
+ status_error = (vm_status_error == "no") and (addition_status_error
== "no")
What is addition_status_error for? I think I asked before, because
it's only 'and' here with status_error. Other people may have same
question, it would be good to put comment here and in config.
explaining what addition_status_error is used for. Sorry, I have bad
memory sometimes.
I'm very sorry, it's my fault. I will add explain as possible as I can.:-P
+ mem_status_error = params.get("mem_status_error", "no")
+ status_error = status_error and (mem_status_error == "no")
+ if status_error:
+ if status != 0:
+ raise error.TestFail("Run failed with right command!")
+ else:
+ if new_memory and current_memory != new_memory:
+ raise error.TestFail(
+ "Run successful but result is not expected")
+ else:
+ if status == 0:
+ raise error.TestFail("Run successful with wrong command")
-- 1.7.1
--
Best Regards
Yu Mingfei
Otherwise looks good. It follows a familiar pattern as other tests
which helps make it easy to understand. I like how the test module is
very simple, and does not try to do complex operations.
Thanks
Yu
_______________________________________________
Autotest-kernel mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/autotest-kernel
