11:01 a.m.
Kaitlin Rupert wrote:
> +sup_types = ['Xen', 'KVM', 'XenFV', 'LXC']
> +slp_path="/usr/bin/slptool"
It's possible that slptool might be installed in a different location.
You don't have to do this, but instead of checking the path, you could
use run_remote() to check whether "slptool help" returns an error. If
it does, slptool isn't installed (or the path is non-standard).
> +
> +def get_slp_info(server):
> +
> + if not os.path.exists(slp_path):
> + logger.error("Slp tool does not exist on the machine ")
> + return SKIP
> +
> + logger.info("Slp tool found on the machine ....")
> +
> + # The test is written to work with Pegasus for now.
> + # In future we can include sfcb support as well
> + # When sfcb support will be planned then replace the following
> check + # with check_cimom() fn of common_util.py lib
> + cmd = "ps -ef | grep -v grep | grep cimserver"
> + rc, out = run_remote(server, cmd)
> + if rc != 0:
> + logger.error("cimserver not found on '%s'", server)
> + return FAIL
I would SKIP here. Also, can you add something like "test only
supported for sfcb at this time" - so the user knows why the test is
skipped.
> +
> + cmd = "cimconfig -l -p | grep slp"
> + rc, out = run_remote(server, cmd)
> + if rc != 0:
> + logger.error("slp is not enabled in the cimserver on '%s'",
server)
> + return FAIL
I would skip here as well.
> +
> + return PASS
> +
> +def get_slp_attrs(server):
> + slp_attrs = None
> + status = FAIL
> +
> + cmd = "%s findsrvs service:wbem | grep -w http | "\
> + "grep -e '%s'" % (slp_path, server)
> + rc, out = run_remote(server, cmd)
> + if rc != 0:
> + logger.error("Failed to find slp service on '%s'", server)
> + return status, slp_attrs
Instead of calling findsrvs, why not call findattrs directly?
I call findsrvs to
get all the ports which could be used to find the attrs.
> +
> + attrs_in = out.rsplit(",")
> + slp_query = attrs_in[0]
> + cmd = "%s findattrs %s" % (slp_path, slp_query)
> + rc, slp_attrs = run_remote(server, cmd)
> + if len(slp_attrs) == 0:
> + # Getting the query and the port details
> + slp_query_info = slp_query.rsplit(":", 1)
> + slp_service_query = slp_query_info[0]
> + port_used = slp_query_info[1]
> + logger.info("Did not get the profile information for '%s'",
slp_query)
> + logger.info("Trying with ports other than '%s'....", port_used)
> +
> + # Remove the already used port information + ports =
> attrs_in.remove(slp_query)
> + for port in attrs_in:
> + new_slp_query = slp_service_query + ":" + port
> + cmd = "%s findattrs %s" % (slp_path, new_slp_query)
> + rc, slp_attrs = run_remote(server, cmd)
> + if len(slp_attrs) != 0: + break
If I force the test to execute this block of code, I get the following
error:
ERROR - ValueError : list.remove(x): x not in list
Traceback (most recent call last):
File "./lib/XenKvmLib/const.py", line 140, in do_try
File "04_verify_libvirt_cim_slp_profiles.py", line 167, in main
status, slp_attrs = get_slp_attrs(ip_addr)
File "04_verify_libvirt_cim_slp_profiles.py", line 96, in get_slp_attrs
ports = attrs_in.remove(slp_query)
ValueError: list.remove(x): x not in list
I think it's okay to remove this bit of code. We have to enable http
connections with the CIMOM in order to run cimtest. If we want cimtest
to work with just https, we'd need to do a lot of cleanup. That means
if the CIMOM is advertising via slp correctly, then the advertisement
will occur with the http port (otherwise its an error with the
advertisement).
You can put a comment here indicating the test only works when the
CIMOM is enabled for http. That's fine for now.
I am not using https at all. I am parsing the port associated with the
http record returned by the slptool findsrvs..
Let me know if I am wrong in understanding what you mean here.
> +
> + if len(slp_attrs) != 0:
> + status = PASS
> +
> + return status, slp_attrs
> +
> +def filter_reg_name_from_slp(slp_attrs):
> + slp_profile_list = []
> +
> + for line in slp_attrs.split('\n'):
> + lines=line.split("RegisteredProfilesSupported")
Spaces around the = needed here.
> + dmtf_profiles=lines[1].split("DMTF")
And here.
> +
> +@do_main(sup_types)
> +def main():
> + options = main.options
> + server = options.ip
> + virt = options.virt
> + status = FAIL
> +
> + # Making sure that the server information passed is + # hostname or
> ip address
> + if server == "localhost":
> + logger.error("Please specify hostname or ip address, than '%s'",
+
> server)
> + return SKIP
Why not use gethostname() to get the hostname?
> + + status = get_slp_info(server)
> + if status != PASS:
> + return status +
> + ip_addr = gethostbyaddr(server)[2][0]
You can use gethostbyname(), which is a little cleaner. You can pass
it the output from gethostname().
I tried supplying the hostname to the slp command, but it does not seem
to work. Hence I choose gethostbyaddr() to get the ip from the hostname.
Let me know if we can access the attrs while passing the hostname.
> + status, slp_attrs = get_slp_attrs(ip_addr)
> + if status != PASS:
> + logger.error("Failed to get slp attributes on %s", server)
> + return status
> +
> + slp_profile_list = filter_reg_name_from_slp(slp_attrs)
> +
> + status, libvirt_cim_reg_list = get_libvirt_cim_profile_info(server,
> virt)
> + if status != PASS:
> + logger.error("Failed to enumerate profile information on %s", server)
> + return status
> +
> + # Make sure all the Libvirt-CIM profiles are advertised via slp
> + if (libvirt_cim_reg_list) <= (slp_profile_list):
> + logger.info("Successfully verified the Libvirt-CIM profiles")
> + return PASS
> + else:
No need for the else here.
Don't we want to print an error message if we dont
get the expected
Libvirt-CIM profiles ?
> + logger.error("Mismatch in the profiles registered")
> + logger.error("Slp returned profile --> %s,\n Libvirt-CIM expected "
> + "profiles %s", slp_profile_list, libvirt_cim_reg_list)
> + return FAIL
> +
> +if __name__=="__main__":
> + sys.exit(main())
> +
>
--
Thanks and Regards,
Deepti B. Kalakeri
IBM Linux Technology Center
deeptik(a)linux.vnet.ibm.com