[PATCH] NetTestController: iterate only over StaticInterfaces to get driver info
by Jan Tluka
This is a fix for corner case bug introduced by recent commit
92b8ae0dc8ac25a10835d9898172980b3af89ec1.
The problem is that if the match results with at least one
UnusedInterface we end up iterating over such interface that
unfortunately does not have self._id set since it defaults to
None.
The solution is to query driver information for StaticInterface
interface classes only.
Fixes #131
Signed-off-by: Jan Tluka <jtluka(a)redhat.com>
---
lnst/Controller/NetTestController.py | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/lnst/Controller/NetTestController.py b/lnst/Controller/NetTestController.py
index e12d349..ec0db54 100644
--- a/lnst/Controller/NetTestController.py
+++ b/lnst/Controller/NetTestController.py
@@ -27,6 +27,7 @@ from lnst.Common.NetTestCommand import str_command, CommandException
from lnst.Controller.RecipeParser import RecipeParser, RecipeError
from lnst.Controller.SlavePool import SlavePool
from lnst.Controller.Machine import MachineError, VirtualInterface
+from lnst.Controller.Machine import StaticInterface
from lnst.Common.ConnectionHandler import send_data, recv_data
from lnst.Common.ConnectionHandler import ConnectionHandler
from lnst.Common.Config import lnst_config
@@ -201,7 +202,8 @@ class NetTestController:
for iface in ifaces:
iface.configure()
- if m._libvirt_domain is None:
+ if (m._libvirt_domain is None and
+ isinstance(iface, StaticInterface)):
driver = iface._driver
if_id = iface._id
mapped_machine = self._slave_pool._map['machines'][m_id]
--
1.9.3
8 years, 9 months
[PATCH v3 00/12] various bug fixes and enhancements
by Ondrej Lichtner
From: Ondrej Lichtner <olichtne(a)redhat.com>
This is a big longer patchset that contains various bug fixes and enhancements
that I've come across while working on LNST-PerfRepo integration. The patche(s)
porting our current ENRT recipes to support PerfRepo will come later when I've
successfully ported all of them.
v3 fixes a merge conflict of the "PerfRepo: log http..." patch
Ondrej Lichtner (12):
NetTestSlave: fix traceback for down of nonexistent device
deconfiguration: restore_system_config before interface deconfiguration
PerfRepo: add get_testUid method
Task: improve generate_hash method
Task: get_baseline returns a TestExecution object instead of a report id
Netperf: small fixes
Netperf: check return code before output parsing
add PerfRepoUtils
Utils: std_deviation return 0 for an empty list
Netperf: avoid division by zero
PerfRepo: log http response on failed request
NetTestSlave: strip contents of kernel_release and redhat_release
lnst/Common/Utils.py | 2 ++
lnst/Controller/Machine.py | 2 ++
lnst/Controller/PerfRepo.py | 13 +++++++++++++
lnst/Controller/PerfRepoUtils.py | 33 +++++++++++++++++++++++++++++++++
lnst/Controller/Task.py | 20 +++++++++++++++-----
lnst/Slave/NetTestSlave.py | 18 +++++++++++++-----
test_modules/Netperf.py | 35 +++++++++++++++++++++++++----------
7 files changed, 103 insertions(+), 20 deletions(-)
create mode 100644 lnst/Controller/PerfRepoUtils.py
--
2.1.0
8 years, 9 months
[PATCH v2 00/12] various bug fixes and enhancements
by Ondrej Lichtner
From: Ondrej Lichtner <olichtne(a)redhat.com>
This is a big longer patchset that contains various bug fixes and enhancements
that I've come across while working on LNST-PerfRepo integration. The patche(s)
porting our current ENRT recipes to support PerfRepo will come later when I've
successfully ported all of them.
Ondrej Lichtner (12):
NetTestSlave: fix traceback for down of nonexistent device
deconfiguration: restore_system_config before interface deconfiguration
PerfRepo: add get_testUid method
Task: improve generate_hash method
Task: get_baseline returns a TestExecution object instead of a report id
Netperf: small fixes
Netperf: check return code before output parsing
add PerfRepoUtils
Utils: std_deviation return 0 for an empty list
Netperf: avoid division by zero
PerfRepo: log http response on failed request
NetTestSlave: strip contents of kernel_release and redhat_release
lnst/Common/Utils.py | 2 ++
lnst/Controller/Machine.py | 2 ++
lnst/Controller/PerfRepo.py | 13 +++++++++++++
lnst/Controller/PerfRepoUtils.py | 33 +++++++++++++++++++++++++++++++++
lnst/Controller/Task.py | 20 +++++++++++++++-----
lnst/Slave/NetTestSlave.py | 18 +++++++++++++-----
test_modules/Netperf.py | 35 +++++++++++++++++++++++++----------
7 files changed, 103 insertions(+), 20 deletions(-)
create mode 100644 lnst/Controller/PerfRepoUtils.py
--
2.1.0
8 years, 9 months
[PATCH 00/12] various bug fixes and enhancements
by Ondrej Lichtner
From: Ondrej Lichtner <olichtne(a)redhat.com>
This is a big longer patchset that contains various bug fixes and enhancements
that I've come across while working on LNST-PerfRepo integration. The patche(s)
porting our current ENRT recipes to support PerfRepo will come later when I've
successfully ported all of them.
Ondrej Lichtner (12):
NetTestSlave: fix traceback for down of nonexistent device
deconfiguration: restore_system_config before interface deconfiguration
PerfRepo: add get_testUid method
Task: improve generate_hash method
Task: get_baseline returns a TestExecution object instead of a report id
Netperf: small fixes
Netperf: check return code before output parsing
add PerfRepoUtils
Utils: std_deviation return 0 for an empty list
Netperf: avoid division by zero
PerfRepo: log http response on failed request
NetTestSlave: strip contents of kernel_release and redhat_release
lnst/Common/Utils.py | 2 ++
lnst/Controller/Machine.py | 2 ++
lnst/Controller/PerfRepo.py | 13 +++++++++++++
lnst/Controller/PerfRepoUtils.py | 33 +++++++++++++++++++++++++++++++++
lnst/Controller/Task.py | 20 +++++++++++++++-----
lnst/Slave/NetTestSlave.py | 18 +++++++++++++-----
test_modules/Netperf.py | 31 +++++++++++++++++++++++--------
7 files changed, 101 insertions(+), 18 deletions(-)
create mode 100644 lnst/Controller/PerfRepoUtils.py
--
2.1.0
8 years, 9 months
[PATCH] regression-tests: fix test #20
by Jan Tluka
This test is supposed to be for case when we do config_only first and
then run of a recipe with team configuration. But lnst-ctl was executed
in run mode instead.
Signed-off-by: Jan Tluka <jtluka(a)redhat.com>
---
regression-tests/tests/20/run.sh | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/regression-tests/tests/20/run.sh b/regression-tests/tests/20/run.sh
index ca50115..8a720f0 100755
--- a/regression-tests/tests/20/run.sh
+++ b/regression-tests/tests/20/run.sh
@@ -4,7 +4,7 @@
init_test
-lnst-ctl -d run recipe1.xml | tee test.log
+lnst-ctl -d config_only recipe1.xml | tee test.log
rv1=${PIPESTATUS[0]}
lnst-ctl -d run recipe2.xml | tee test.log
--
1.9.3
8 years, 9 months