Adding iperf test
PATCH 1/1 Adding iperf test
Adding iperf network performance test. Basically it tests networking functionality, stability and performance of guest OSes. This test is cross-platform -- i.e. it works on both Linux and Windows VMs.
--- client/tests/kvm/kvm.py | 1 + client/tests/kvm/kvm_iperf.py | 105 +++++++++++++++++++++++++++++++++ client/tests/kvm/kvm_tests.cfg.sample | 8 +++ 3 files changed, 114 insertions(+), 0 deletions(-) create mode 100644 client/tests/kvm/kvm_iperf.py
client/tests/kvm/kvm.py +1 -0
@@ -53,6 +53,7 @@ class kvm(test.test): "autotest": test_routine("kvm_tests", "run_autotest"), "kvm_install": test_routine("kvm_install", "run_kvm_install"), "linux_s3": test_routine("kvm_tests", "run_linux_s3"), + "iperf": test_routine("kvm_iperf", "run_iperf"), } # Make it possible to import modules from the test's bindir
client/tests/kvm/kvm_iperf.py new file +105 -0
105 lines not on this page · raw message
client/tests/kvm/kvm_tests.cfg.sample +8 -0
@@ -82,6 +82,10 @@ variants: - linux_s3: install setup type = linux_s3 + - iperf: install setup + type = iperf + extra_params += " -snapshot" + # NICs variants: - @rtl8139:
@@ -102,6 +106,8 @@ variants: ssh_status_test_command = echo $? username = root password = 123456 + iperf: + iperf_binary = misc/iperf variants: - Fedora:
@@ -292,6 +298,8 @@ variants: password = 123456 migrate: migration_test_command = ver && vol + iperf: + iperf_binary = misc/iperf.exe variants: - Win2000:
Signature · 1 line
1.6.2.2
Quoted 4 lines · Lucas Meneghel Rodrigues
Adding iperf network performance test. Basically it tests networking functionality, stability and performance of guest OSes. This test is cross-platform -- i.e. it works on both Linux and Windows VMs.
Ok, now that I had rebased the test, I have a few comments to say:
* I don't like the idea to ship binaries inside the test. Fair enough
that we don't support other archs than x86 and x86_64 (also there's the problem windows usually doesn't ship a working toolchain), but I would like to think a bit more about it.
* Autotest already does have an iperf test, that could be used on Linux
guests. Sure there's a problem of matching what's being executed on windows, but it's worth a look
* Autotest iperf test usually runs with 2 machines, one in 'server'
role and other in 'client' mode. I would like to pursue the same model, 2 vms, one running as a server and another as a client.
Alexey and Yaniv, I'd like to hear your opinions on this.
Thanks,
Quoted 164 lines
Signed-off-by Alexey Eromenko <[email protected]>--- client/tests/kvm/kvm.py | 1 + client/tests/kvm/kvm_iperf.py | 105 +++++++++++++++++++++++++++++++++ client/tests/kvm/kvm_tests.cfg.sample | 8 +++ 3 files changed, 114 insertions(+), 0 deletions(-) create mode 100644 client/tests/kvm/kvm_iperf.pyclient/tests/kvm/kvm.py +1
-0@@ -53,6 +53,7 @@ class kvm(test.test): "autotest": test_routine("kvm_tests", "run_autotest"), "kvm_install": test_routine("kvm_install", "run_kvm_install"), "linux_s3": test_routine("kvm_tests", "run_linux_s3"), + "iperf": test_routine("kvm_iperf", "run_iperf"), } # Make it possible to import modules from the test's bindirclient/tests/kvm/kvm_iperf.py new file +105
-0@@ -0,0 +1,105 @@ +import time, os, logging +from autotest_lib.client.common_lib import utils, error +import kvm_utils + +def run_iperf(test, params, env): + """ + Runs iperf on the guest system and brings back the result. + + @see: http://sourceforge.net/projects/iperf + @param test: kvm test object + @param params: Dictionary with test parameters + @param env: Test environment + """ + vm = kvm_utils.env_get_vm(env, params.get("main_vm")) + if not vm: + message = "VM object not found in environment" + logging.error(message) + raise error.TestError, message + if not vm.is_alive(): + message = "VM seems to be dead; Test requires a living VM" + logging.error(message) + raise error.TestError(message) + + logging.info("Waiting for guest to be up...") + + session = kvm_utils.wait_for(vm.ssh_login, 240, 0, 2) + if not session: + message = "Could not log into guest" + logging.error(message) + raise error.TestFail, message + + logging.info("Logged in") + + # Checking for GuestOS-compatible iPerf binary existence on host. + iperf_binary = params.get("iperf_binary", "misc/iperf") + iperf_duration = params.get("iperf_duration", 5) + iperf_parallel_threads = params.get("iperf_parallel_threads", 1) + iperf_dest_ip = params.get("iperf_dest_ip", "10.0.2.2") + iperf_binary = os.path.join(test.bindir, iperf_binary) + if not os.path.exists(iperf_binary): + message = "iPerf binary: %s was not found on host" % iperf_binary + logging.error(message) + raise error.TestError, message + else: + logging.info("iPerf binary: %s was found on host" % iperf_binary) + + # Starting HostOS-compatible iPerf Server on host + logging.info('VM is up ... \n starting iPerf Server on host') + kvm_utils.run_bg("iperf -s", timeout=5) + + # Detecting GuestOS + if iperf_binary.__contains__("exe"): + vm_type="win32" + else: + vm_type="linux32" + + # Copying GuestOS-compatible iPerf binary to guest. + # Starting iPerf Client on guest, plus connect to host. + if vm_type == "win32": + win_dir = "/cygdrive/c/" + logging.info('starting copying %s to Windows VM to %s' % (iperf_binary, + win_dir)) + if not vm.scp_to_remote(iperf_binary, win_dir): + message = "Could not copy Win32 iPerf to guest" + logging.error(message) + raise error.TestError(message) + logging.debug("Enabling file permissions of iPerf.exe on Windows VM...") + session.sendline('cacls C:\iperf.exe /P Administrator:F') + session.sendline('y') + session.sendline('') + time.sleep(2) + session.sendline('') + logging.info("starting iPerf client on Windows VM, connecting to host") + session.sendline('C:\iperf -t %s -c %s -P %s' % (int(iperf_duration), + iperf_dest_ip, + int(iperf_parallel_threads))) + else: + logging.info('starting copying %s to Linux VM ' % iperf_binary) + if not vm.scp_to_remote(iperf_binary, "/usr/local/bin"): + message = "Could not copy Linux iPerf to guest" + logging.error(message) + raise error.TestError, message + print "starting iPerf client on VM, connecting to host" + session.sendline('iperf -t %s -c %s -P %s' % (int(iperf_duration), + iperf_dest_ip, + int(iperf_parallel_threads))) + + # Analyzing results + iperf_result_match, iperf_result = session.read_up_to_prompt() + logging.debug("iperf_result =", iperf_result) + + if iperf_result.__contains__(" 0.00 bits/sec"): + msg = 'Guest returned 0.00 bits/sec during iperf test.' + raise error.TestError(msg) + elif iperf_result.__contains__("No route to host"): + msg = 'SSH to guest returned: No route to host.' + raise error.TestError(msg) + elif iperf_result.__contains__("Access is denied"): + msg = 'SSH to guest returned: Access is denied.' + raise error.TestError(msg) + elif not iperf_result.__contains__("bits/sec"): + msg = 'SSH result unrecognizeable.' + raise error.TestError(msg) + + session.close()client/tests/kvm/kvm_tests.cfg.sample +8
-0@@ -82,6 +82,10 @@ variants: - linux_s3: install setup type = linux_s3 + - iperf: install setup + type = iperf + extra_params += " -snapshot" + # NICs variants: - @rtl8139:@@ -102,6 +106,8 @@ variants: ssh_status_test_command = echo $? username = root password = 123456 + iperf: + iperf_binary = misc/iperf variants: - Fedora:@@ -292,6 +298,8 @@ variants: password = 123456 migrate: migration_test_command = ver && vol + iperf: + iperf_binary = misc/iperf.exe variants: - Win2000:
Signature · 3 lines
Lucas Meneghel Rodrigues Software Engineer (QE) Red Hat - Emerging Technologies
Quoted 3 lines · Lucas Meneghel Rodrigues
* Autotest iperf test usually runs with 2 machines, one in 'server'
role and other in 'client' mode. I would like to pursue the same model, 2 vms, one running as a server and another as a client.
Nevermind this comment, I've missed the part of the test that starts the server on the host linux system.
Signature · 3 lines
Lucas Meneghel Rodrigues Software Engineer (QE) Red Hat - Emerging Technologies
PATCH 1/1 bugfix for changing kvm_log to logging
In reply to Lucas Meneghel Rodriguesclient/tests/kvm/kvm.py +1 -1
@@ -73,7 +73,7 @@ class kvm(test.test): resource.setrlimit(resource.RLIMIT_CORE, (-1, -1)) # Set the logging prefix - #kvm_log.set_prefix(params.get("shortname")) + #logging.set_prefix(params.get("shortname")) # Report the parameters we've received and write them as keyvals logging.debug("Test parameters:")
client/tests/kvm/kvm_utils.py +1 -1
@@ -670,7 +670,7 @@ def remote_login(command, password, prompt, linesep="\n", timeout=10): sub.close() return None elif match == 4: # "Connection refused" - kvm_log.debug("Got 'Connection refused'") + logging.debug("Got 'Connection refused'") sub.close() return None elif match == 5: # prompt
Lucas Meneghel Rodrigues
AppliedRe: [KVM-AUTOTEST PATCH] bugfix for changing kvm_log to logging
Applied, with minimum changes
client/tests/kvm/kvm.py class kvm(test.test): · line 76
Quoted 10 lines · Yolkfull Chow
client/tests/kvm/kvm.py +1
-1@@ -73,7 +73,7 @@ class kvm(test.test): resource.setrlimit(resource.RLIMIT_CORE, (-1, -1)) # Set the logging prefix - #kvm_log.set_prefix(params.get("shortname")) + #logging.set_prefix(params.get("shortname"))
This line was commented, the logging module does not have a set_prefix function. Therefore, the line was just removed.
Quoted 16 lines
client/tests/kvm/kvm.py +0
-0# Report the parameters we've received and write them as keyvals logging.debug("Test parameters:")client/tests/kvm/kvm_utils.py +0
-0@@ -670,7 +670,7 @@ def remote_login(command, password, prompt,linesep="\n", timeout=10):
sub.close() return None elif match == 4: # "Connection refused"client/tests/kvm/kvm_utils.py +1
-1- kvm_log.debug("Got 'Connection refused'") + logging.debug("Got 'Connection refused'") sub.close() return None elif match == 5: # prompt
Quoted 5 lines · Lucas Meneghel Rodrigues
Adding iperf network performance test. Basically it tests networking functionality, stability and performance of guest OSes. This test is cross-platform -- i.e. it works on both Linux and Windows VMs.
I have a question here. Why are we adding iperf in a way different than other tests ? We have client/tests/<different_tests> directory for each test which contains the python modules and the test tarball. Then why in case of iperf we are putting it under client/tests/kvm and modifying kvm.py instead of putting the testsuit as part of autotest(run_autotest is not enough?)? Even if we do not want to touch the existing iperf test in autotest we can use a separate name like kvm_iperf. Somehow I have a feeling that there was a discussion on the list for keeping tests under a particular directory. But still I feel that should be only for tests specific to KVM and not the guest. Is there any disadvantage of using the current approach of executing these testsuits ?
Quoted 172 lines
Signed-off-by Alexey Eromenko <[email protected]>---
client/tests/kvm/kvm.py | 1 +client/tests/kvm/kvm_iperf.py| 105 +++++++++++++++++++++++++++++++++ client/tests/kvm/kvm_tests.cfg.sample | 8 +++3 files changed, 114 insertions(+), 0 deletions(-)create mode 100644
client/tests/kvm/kvm_iperf.pyclient/tests/kvm/kvm.py +1
-0@@ -53,6 +53,7 @@ class kvm(test.test): "autotest": test_routine("kvm_tests", "run_autotest"), "kvm_install": test_routine("kvm_install", "run_kvm_install"), "linux_s3": test_routine("kvm_tests", "run_linux_s3"), + "iperf": test_routine("kvm_iperf", "run_iperf"), } # Make it possible to import modules from the test's bindirclient/tests/kvm/kvm_iperf.py new file +105
-0@@ -0,0 +1,105 @@ +import time, os, logging +from autotest_lib.client.common_lib import utils, error +import kvm_utils + +def run_iperf(test, params, env): + """ + Runs iperf on the guest system and brings back the result. + + @see: http://sourceforge.net/projects/iperf + @param test: kvm test object + @param params: Dictionary with test parameters + @param env: Test environment + """ + vm = kvm_utils.env_get_vm(env, params.get("main_vm")) + if not vm: + message = "VM object not found in environment" + logging.error(message) + raise error.TestError, message + if not vm.is_alive(): + message = "VM seems to be dead; Test requires a living VM" + logging.error(message) + raise error.TestError(message) + + logging.info("Waiting for guest to be up...") + + session = kvm_utils.wait_for(vm.ssh_login, 240, 0, 2) + if not session: + message = "Could not log into guest" + logging.error(message) + raise error.TestFail, message + + logging.info("Logged in") + + # Checking for GuestOS-compatible iPerf binary existence on host. + iperf_binary = params.get("iperf_binary", "misc/iperf") + iperf_duration = params.get("iperf_duration", 5) + iperf_parallel_threads = params.get("iperf_parallel_threads", 1) + iperf_dest_ip = params.get("iperf_dest_ip", "10.0.2.2") + iperf_binary = os.path.join(test.bindir, iperf_binary) + if not os.path.exists(iperf_binary): + message = "iPerf binary: %s was not found on host" % iperf_binary + logging.error(message) + raise error.TestError, message + else: + logging.info("iPerf binary: %s was found on host" % iperf_binary) + + # Starting HostOS-compatible iPerf Server on host + logging.info('VM is up ... \n starting iPerf Server on host') + kvm_utils.run_bg("iperf -s", timeout=5) + + # Detecting GuestOS + if iperf_binary.__contains__("exe"): + vm_type="win32" + else: + vm_type="linux32" + + # Copying GuestOS-compatible iPerf binary to guest. + # Starting iPerf Client on guest, plus connect to host. + if vm_type == "win32": + win_dir = "/cygdrive/c/" + logging.info('starting copying %s to Windows VM to %s' % (iperf_binary, + win_dir)) + if not vm.scp_to_remote(iperf_binary, win_dir): + message = "Could not copy Win32 iPerf to guest" + logging.error(message) + raise error.TestError(message) + logging.debug("Enabling file permissions of iPerf.exe on Windows VM...") + session.sendline('cacls C:\iperf.exe /P Administrator:F') + session.sendline('y') + session.sendline('') + time.sleep(2) + session.sendline('') + logging.info("starting iPerf client on Windows VM, connecting to host") + session.sendline('C:\iperf -t %s -c %s -P %s' % (int(iperf_duration), + iperf_dest_ip, + int(iperf_parallel_threads))) + else: + logging.info('starting copying %s to Linux VM ' % iperf_binary) + if not vm.scp_to_remote(iperf_binary, "/usr/local/bin"): + message = "Could not copy Linux iPerf to guest" + logging.error(message) + raise error.TestError, message + print "starting iPerf client on VM, connecting to host" + session.sendline('iperf -t %s -c %s -P %s' % (int(iperf_duration), + iperf_dest_ip, + int(iperf_parallel_threads))) + + # Analyzing results + iperf_result_match, iperf_result = session.read_up_to_prompt() + logging.debug("iperf_result =", iperf_result) + + if iperf_result.__contains__(" 0.00 bits/sec"): + msg = 'Guest returned 0.00 bits/sec during iperf test.' + raise error.TestError(msg) + elif iperf_result.__contains__("No route to host"): + msg = 'SSH to guest returned: No route to host.' + raise error.TestError(msg) + elif iperf_result.__contains__("Access is denied"): + msg = 'SSH to guest returned: Access is denied.' + raise error.TestError(msg) + elif not iperf_result.__contains__("bits/sec"): + msg = 'SSH result unrecognizeable.' + raise error.TestError(msg) + + session.close()client/tests/kvm/kvm_tests.cfg.sample +0
-0@@ -82,6 +82,10 @@ variants:-
linux_s3: install setup
type =linux_s3+ - iperf: install setup
+ type = iperf
+ extra_params += " -snapshot"
+
# NICs
variants:
- @rtl8139:client/tests/kvm/kvm_tests.cfg.sample +0
-0@@ -102,6 +106,8 @@ variants:
ssh_status_test_command= echo $?
username = root
password = 123456
+ iperf:
+ iperf_binary = misc/iperfvariants:
- Fedora:client/tests/kvm/kvm_tests.cfg.sample +0
-0@@ -292,6 +298,8 @@ variants:password = 123456
migrate:
migration_test_command= ver && vol
+ iperf:
+ iperf_binary = misc/iperf.exevariants:
- Win2000:
--
1.6.2.2--
To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Signature · 1 line
Sudhir Kumar
Quoted 7 lines · sudhir kumar
Quoted 5 lines · Lucas Meneghel Rodrigues
Adding iperf network performance test. Basically it tests networking functionality, stability and performance of guest OSes. This test is cross-platform -- i.e. it works on both Linux and Windows VMs.
Quoted 12 lines
I have a question here. Why are we adding iperf in a way different than other tests ? We have client/tests/<different_tests> directory for each test which contains the python modules and the test tarball. Then why in case of iperf we are putting it under client/tests/kvm and modifying
kvm.pyinstead of putting the testsuit as part ofautotest(run_autotest is not enough?)? Even if we do not want to touch the existing iperf test in autotest we can use a separate name likekvm_iperf. Somehow I have a feeling that there was a discussion on the list for keeping tests under a particular directory. But still I feel that should be only for tests specific to KVM and not the guest. Is there any disadvantage of using the current approach of executing these testsuits ?
Since the kvm subtests are contained under the kvm test dir, adding the kvm_ file prefixes to the subtests is not necessary IMHO.
Using the autotest iperf test is doable for linux guests, though it doesn't work for windows guests, that's why Alexey decided to implement it from scratch.
Lucas
Quoted 7 lines · sudhir kumar
Quoted 5 lines · Lucas Meneghel Rodrigues
Adding iperf network performance test. Basically it tests networking functionality, stability and performance of guest OSes. This test is cross-platform -- i.e. it works on both Linux and Windows VMs.
Quoted 12 lines
I have a question here. Why are we adding iperf in a way different than other tests ? We have client/tests/<different_tests> directory for each test which contains the python modules and the test tarball. Then why in case of iperf we are putting it under client/tests/kvm and modifying
kvm.pyinstead of putting the testsuit as part ofautotest(run_autotest is not enough?)? Even if we do not want to touch the existing iperf test in autotest we can use a separate name likekvm_iperf. Somehow I have a feeling that there was a discussion on the list for keeping tests under a particular directory. But still I feel that should be only for tests specific to KVM and not the guest. Is there any disadvantage of using the current approach of executing these testsuits ?
The reason to put my test under "kvm/" test, is because it depends on KVM-Autotest framework, not just on generic Autotest framework.
In addition, the test is cross-platform on the guest side, currently supporting Windows and Linux guests, with possibility to support Solaris and BSD in future.
LMR: me too, hate putting binaries in source tree, but the alternative option is to provide separate *.tar.bz2 for all the binary utils, and I don't sure which way is better.
Signature · 1 line
-Alexey Eromenko
Quoted 3 lines · Alexey Eremenko
LMR: me too, hate putting binaries in source tree, but the alternative option is to provide separate *.tar.bz2 for all the binary utils, and I don't sure which way is better.
Yes, I don't have a clear idea as well. It's currently under discussion...
Lucas
Quoted 7 lines · Lucas Meneghel Rodrigues
Quoted 3 lines · Alexey Eremenko
LMR: me too, hate putting binaries in source tree, but the alternative option is to provide separate *.tar.bz2 for all the binary utils, and I don't sure which way is better.
Yes, I don't have a clear idea as well. It's currently under discussion...
Is KVM x86_64 only?
Quoted 8 lines · Martin Bligh
Quoted 6 lines · Lucas Meneghel Rodrigues
Quoted 4 lines · Alexey Eremenko
LMR: me too, hate putting binaries in source tree, but the alternative option is to provide separate *.tar.bz2 for all the binary utils, and I don't sure which way is better.
Quoted 6 lines
Quoted 3 lines
Yes, I don't have a clear idea as well. It's currently under discussion...
Is KVM
x86_64only?
It's x86-64, i386, ia64, s390, and powerpc 44x/e500 only.
Signature · 1 line
error compiling committee.c: too many arguments to function
Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
Quoted 4 lines
LMR: me too, hate putting binaries in source tree, but the alternative option is to provide separate *.tar.bz2 for all the binary utils, and I don't sure which way is better.
Quoted 8 lines
Quoted 6 lines
Quoted 3 lines
Yes, I don't have a clear idea as well. It's currently under discussion...
Is KVM
x86_64only?It's x86-64, i386, ia64, s390, and powerpc 44x/e500 only.
OK, then it's difficult to see using binaries? Can we not compile these on the system at use time (see the client/deps directory for other stuff we do this for)
M.
Quoted 9 lines · Martin Bligh
Quoted 9 lines
Quoted 9 lines
Quoted 8 lines
Quoted 4 lines
LMR: me too, hate putting binaries in source tree, but the alternative option is to provide separate *.tar.bz2 for all the binary utils, and I don't sure which way is better.
Yes, I don't have a clear idea as well. It's currently under discussion...
Quoted 8 lines
Quoted 4 lines
Quoted 2 lines
Is KVM
x86_64only?It's x86-64, i386, ia64, s390, and powerpc 44x/e500 only.
OK, then it's difficult to see using binaries? Can we not compile these on the system at use time (see the client/deps directory for other stuff we do this for)
Biggest trouble is compiling the test under windows hosts. We are figuring out a way to work around this problem.
Quoted 11 lines · Martin Bligh
Quoted 5 lines
Quoted 3 lines
Is KVM
x86_64only?It's x86-64, i386, ia64, s390, and powerpc 44x/e500 only.
OK, then it's difficult to see using binaries? Can we not compile these on the system at use time (see the client/deps directory for other stuff we do this for)
note kvm-autotest is x86 only at this time.
Signature · 1 line
error compiling committee.c: too many arguments to function