Age | Commit message (Collapse) | Author | Files | Lines |
|
Workaround for "ZeroDivisionError: float division by zero" and
"RuntimeError: Unable to retrieve TSC" issues.
The actual clean fix requires reworking get_data() to accumulate
bytes received from PROX TCP port until EOL is received.
Since the other functions calling get_data() expect a synchronous
answer from PROX, timeout should be considered a fatal error, and
the current default at 10ms was just calling for problems!
Even if this change does not get merged, at least this workaround
will be available to end-users for cherry-picking from gerrit,
while the actual clean fix is being implemented.
JIRA: YARDSTICK-1381
Change-Id: I6af226ca31aaebeb50302e24b1eb209de70f6526
Signed-off-by: Patrice Buriez <patrice.buriez@intel.com>
|
|
JIRA: YARDSTICK-1591
Change-Id: I3ea9039d25bfce578681adb9e27e1598e84a1f56
Signed-off-by: John O Loughlin <john.oloughlin@intel.com>
|
|
The existing implementation of IPC is not finished and isn't used
by any of the VNFs/TG implementation. It is used in the code but does
nothing from functionality perspective.
New syncronization mechanism is going to be implemented by using
different approach than it was designed before. Thus, the current
IPC mechanism is not going to be re-used. So, removing it.
The IPC consumer/producer implementation is left as it may be
required for other purposes.
- Remove SampleVNF MQ consumer class
- Remove IterationIPC MQ producer for VNF control messages
- Remove MQ producer from SampleVNFTrafficGen class
- Remove TrafficGeneratorProducer class
- Remove IterationIPC runner
- Remove unused task_id form VNF Generic initialization as it is not
required for synchronization of VNFs/TGs anymore.
- Fix UT
JIRA: YARDSTICK-1592
Change-Id: I65fe51bcbd1bfeea0c43eb79ca6fb2aab5b65ae7
Signed-off-by: Mytnyk, Volodymyr <volodymyrx.mytnyk@intel.com>
|
|
JIRA: YARDSTICK-1571
Cores in PROX support multiple configurations
NSB supports today the more basic one, i.e. [core x]
When one wants to use multiple cores sharing the same
configuration, instead of copying the whole [core]
section, PROX support the following syntax
[core x-y]
where x is the 1st core and y the last one
or (for instance) [core x,y,z]
NSB Now supports
[core a,b,c,d]
[core a-d]
[core a,c-d]
Change-Id: I34cd107143c89e16d58e7a99e1887ffbf720a5d1
Signed-off-by: Daniel Martin Buckley <daniel.m.buckley@intel.com>
|
|
|
|
JIRA: YARDSTICK-1524
Add dashboard, Traffic Profile support, PROX IRQ support, topology
Change-Id: I0c0cde66f4f4d3d25aa77c502bd4349cc1b3eca2
Signed-off-by: Daniel Martin Buckley <daniel.m.buckley@intel.com>
|
|
historical data
JIRA: YARDSTICK-1458
The TG is only sampled at the end of every test interval.
This is incorrect. It should be sampled every interval.
Change-Id: I612ead8a243a7d9930874d02adc9b959b843b99b
Signed-off-by: Daniel Martin Buckley <daniel.m.buckley@intel.com>
|
|
JIRA: YARDSTICK-1458
Removed excessive jitter by removing extra sleeps in start/stop/reset
ssh
Removed extra retries when bad data returned
Added a slep at end of test to give runner a chance to empty the queue
on shutdown
Added a timeout on port retry attempts
Change-Id: Ica60b11c7af242817a83c154157827581f60d16a
Signed-off-by: Daniel Martin Buckley <daniel.m.buckley@intel.com>
|
|
JIRA: YARDSTICK-1419
When using prox_binsearch algorithm, a binary search is
performed, increasing lower bound when step was successful
and decreasing upper bound when the step was a failure.
This runs until the test_precision (as specified in the
traffic profile) is reached. When the test precision
is reached, the test is not completed until the
runner duration is reached.
As runner duration is usually high (e.g. 1800sec), the
tests take much too long to execute.
This makes it difficult to create test suites.
Change-Id: I6cc503a09fb534a556c61c805e6df4786bb8cc4b
Signed-off-by: Daniel Martin Buckley <daniel.m.buckley@intel.com>
|
|
Now all traffic generators using the default implementation of
"ClientResourceHelper.run_traffic" will update the status of
the traffic injection using the MQ.
The available methods are listed in common.messsaging (VNF_METHOD_*):
- tg_method_started: VNF app started
- tg_method_finished: VNF app finished
- tg_method_iteration: VNF app execution loop started
JIRA: YARDSTICK-1127
Change-Id: I12829e1762fc20cc95da3b50767a66f031e25ee8
Signed-off-by: Rodolfo Alonso Hernandez <rodolfo.alonso.hernandez@intel.com>
|
|
When setting interface_speed option in the test case, NSB/PROX
was still considering 100% as 10 Gbps.
With this fix, when interface_speed is set to for instance 25 Gbps
then NSB/PROX will start generate at 25 Gbps.
Yardstick logging info printed on the screen will show performance result
as percentage of interface_speed (e.g. 100% will be 25 Gbps).
This is done by sending the command "speed 250" to PROX (PROX cannot
know the underlying interface speed in all cases e.g. in virtio, and
hence considers 100% as 10Gbps).
JIRA: YARDSTICK-1217
Change-Id: I4989161b04a7afa5e347935ee7ccef7dfe5a504d
Signed-off-by: Xavier Simonart <xavier.simonart@intel.com>
Signed-off-by: Emma Foley <emma.l.foley@intel.com>
|
|
|
|
JIRA: YARDSTICK-1244
When sending the start or stop command to PROX. We sould not send the same core.
We should not send "start 1,1" instead we should send "start 1"
Change-Id: Ie600dc3ed808fd00341e92f13bd52199e68dc51f
Signed-off-by: Daniel Martin Buckley <daniel.m.buckley@intel.com>
|
|
JIRA: YARDSTICK-1219
Currently NSB for NFVI using PROX returns sampling information every 13
seconds.
This is too slow.
It is required to return sampling information form Generator AND VNF at
least every 1 second.
This change is depandant on JIRA: YARDSTICK-1212 and YARDSTICK-1220
Change-Id: Ica7ab795a2919d191d8cd846d028e15739e33fb7
Signed-off-by: Daniel Martin Buckley <daniel.m.buckley@intel.com>
|
|
JIRA: YARDSTICK-1212
Change-Id: Ic4d4a3c00e4e278f4de06cc176ff663892895569
Signed-off-by: Xavier Simonart <xavier.simonart@intel.com>
|
|
|
|
JIRA: YARDSTICK-1037
NSB PROX L3FWD was incorrect. The routing table (ipv4.lua) was incorrect.
Also a specific ipv4 for 2 port configuration is required.
Change-Id: I3a5fd1ea1d1ddf74cbdd6a5d9e7ced42d3c1d6e9
Signed-off-by: Daniel Martin Buckley <daniel.m.buckley@intel.com>
Signed-off-by: Abhijit Sinha <abhijit.sinha@intel.com>
|
|
JIRA: YARDSTICK-1035
Do not hardcode NIC/interface speed in PROX test. Test assumes
NIC used is 10Gbps. This is incorrect. It could support 1Gbps,
10Gbps, 25Gbps, 40Gbps or something else.
This is used to calculate pps (Packets Per Second)
In Baremetal the NIC speed could be extracted. however when
run on a virtual machine this is not possible.
Solution:
Add in options section of test file.
eg.
Options:
interface_speed_gbps: 10
Where 10 refers to a 10Gbps. In a setup where multiple interfaces
are used. This will refer to the speed of the slowest connection.
Change-Id: I89ab16479a2cdd1d79e52cbcc5a972762c60d057
Signed-off-by: Daniel Martin Buckley <daniel.m.buckley@intel.com>
|
|
"traffic_profile" modules should be imported only once. Every time
TrafficProfile.get is called, the modules under
"yardstick.network_services.traffic_profiles" are loaded [1]. Instead
of this, the modules should be registered only once the first time
"yardstick.network_services.traffic_profiles.base" is loaded. This
will reduce the execution time and will avoid unnecessary calls.
[1] https://github.com/opnfv/yardstick/blob/d2c7cc4e9768ed003257a95c92cdb278d516761b/yardstick/network_services/traffic_profile/base.py#L36-L37
JIRA: YARDSTICK-951
Change-Id: Ia3565378ba3a1377fcb0aea8bda50ef8189414fd
Signed-off-by: Rodolfo Alonso Hernandez <rodolfo.alonso.hernandez@intel.com>
|
|
JIRA: YARDSTICK-1036
This stores a number of extra counters in influxdb for Prox test cases.
It also stores existing counters with a "succcess_" tag.
Previously throughput where stored without success or failure indication.
Also "Result_" counters are also stored.
These can now be used by Grafana to graph output.
Change-Id: Ie5636c14ecbab1b53a988bdfbd47ddd1fcdbd695
Signed-off-by: Daniel Martin Buckley <daniel.m.buckley@intel.com>
|
|
Tool provisioning in PROX setup environment helper is not needed [1]. The
tool (PROX traffic injector) is already provided during the VNF building:
./ansible/nsb_setup.yml -->
./ansible/build_yardstick_image.yml -->
./ansible/ubuntu_server_cloudimg_modify_samplevnfs.yml -->
./ansible/roles/install_samplevnf
[1]https://github.com/opnfv/yardstick/blob/master/yardstick/network_services/vnf_generic/vnf/prox_helpers.py#L641
JIRA: YARDSTICK-872
Change-Id: I0f925a7967a35a97901fbe5053793a791a7b1b01
Signed-off-by: Rodolfo Alonso Hernandez <rodolfo.alonso.hernandez@intel.com>
|
|
network_services.vnf_generic.vnf.prox_helpers.ProxSocketHelper.rx_stats"
|
|
* Remove the method which is unused and is marked as deprecated.
Change-Id: Ie64084fcd26985283c664445b173a757d3d908ab
JIRA: YARDSTICK-838
Signed-off-by: Emma Foley <emma.l.foley@intel.com>
|
|
JIRA: YARDSTICK-802
Addition of PROX LW_AFTR basked on PROX/DATS v037 test_104_lw_aftr.py
- This support BM and Openstack Heat
- This supports 4 Ports ONLY
- Grafana Dashboards included
- Code Coverage / Unit testing
Change-Id: If2170ab458bf687256d5f1a1e840a3b9d2788ef7
Signed-off-by: Daniel MArtin Buckley <daniel.m.buckley@intel.com>
Signed-off-by: Abhijit Sinha <abhijit.sinha@intel.com>
|
|
Change-Id: I1f457c9c24f2ca84dde61b64f58edaff8952670a
Signed-off-by: Edward MacGillivray <edward.s.macgillivray@intel.com>
|
|
JIRA: YARDSTICK-802
Addition of Prox vPE test case
- The tests supports BM, Openstack Heat
- Supports 4 ports
- Grafana dashboards included
- Added support for parameters.lua
for prox additional files
- Unit tests for code coverage
Change-Id: I5cccb351dacba88a293ae4b8aba1f0a803d62e6d
Signed-off-by: Abhijit Sinha <abhijit.sinha@intel.com>
Signed-off-by: Daniel MArtin Buckley <daniel.m.buckley@intel.com>
Signed-off-by: Ross Brattain <ross.b.brattain@intel.com>
|
|
JIRA: YARDSTICK-802
- Updated the BNG code with minor refactor.
- Corrected the CPE core name
- Update binsearch traffic profile with 64B
Change-Id: Iae0be766edb986520045655fa567651711813a8b
Signed-off-by: Abhijit Sinha <abhijit.sinha@intel.com>
|
|
The PROX tests were hanging in the duration
runner.
These are fixes for various errors:
raise error in collect_kpi if VNF is down
move prox dpdk_rebind after collectd stop
fix dpdk nicbind rebind to group by drivers
prox: raise error in collect_kpi if the VNF is down
prox: add VNF_TYPE for consistency
sample_vnf: debug and fix kill_vnf
pkill is not matching some executable names,
add some debug process dumps and try switching
back to killall until we can find the issue
sample_vnf: add default timeout, so we can override
default 3600 SSH timeout
collect_kpi is the point at which we check
the VNFs and TGs for failures or exits
queues are the problem make sure we aren't silently blocking on
non-empty queues by canceling join thread in subprocess
fixup duration runner to close queues
and other attempt to stop duration runner
from hanging
VnfdHelper: memoize port_num
resource: fail if ssh can't connect
at the end of 3600 second test our ssh connection
is dead, so we can't actually stop collectd
unless we reconnect
fix stop() logic to ignore ssh errors
Change-Id: I6c8e682a80cb9d00362e2fef4a46df080f304e55
Signed-off-by: Ross Brattain <ross.b.brattain@intel.com>
|
|
In some cases we are blocking in base.Runner join() because the
queues are not empty
call cancel_join_thread to prevent the Queue from blocking the
Process exit
https://docs.python.org/3.3/library/multiprocessing.html#all-platforms
Joining processes that use queues
Bear in mind that a process that has put items in a queue will wait
before terminating until all the buffered items are fed by the
"feeder" thread to the underlying pipe. (The child process can call
the cancel_join_thread() method of the queue to avoid this behaviour.)
This means that whenever you use a queue you need to make sure that
all items which have been put on the queue will eventually be removed
before the process is joined. Otherwise you cannot be sure that
processes which have put items on the queue will terminate. Remember
also that non-daemonic processes will be joined automatically.
Warning
As mentioned above, if a child process has put items on a queue (and
it has not used JoinableQueue.cancel_join_thread), then that process
will not terminate until all buffered items have been flushed to the
pipe.
This means that if you try joining that process you may get a deadlock
unless you are sure that all items which have been put on the queue
have been consumed. Similarly, if the child process is non-daemonic
then the parent process may hang on exit when it tries to join all its
non-daemonic children.
cancel_join_thread()
Prevent join_thread() from blocking. In particular, this prevents the
background thread from being joined automatically when the process
exits – see join_thread().
A better name for this method might be allow_exit_without_flush(). It
is likely to cause enqueued data to lost, and you almost certainly
will not need to use it. It is really only there if you need the
current process to exit immediately without waiting to flush enqueued
data to the underlying pipe, and you don’t care about lost data.
Change-Id: I61f11a3b01109d96b7a5445c60f1e171401157fc
Signed-off-by: Ross Brattain <ross.b.brattain@intel.com>
|
|
JIRA: YARDSTICK-802
Added Prox BNG and BNG-QoS Test
- The tests supports BM, Openstack Heat
- Supports 4 ports
- Test added for BNG traffic profile
- Fixed the Prox heat test cases with
proper upstream and downstream links
- Grafana Dashboard for BNG & BNG-QoS added
- Increased the test Duration to 300
TODO:
- Test does not Terminate correctly
Update:
Added new helper class for run_test: Genric, MPLS
and BNG tests.
Change-Id: Ib40811bedb45a3c3030643943f32679a4044e076
Signed-off-by: Abhijit Sinha <abhijit.sinha@intel.com>
Signed-off-by: Edward MacGillivray <edward.s.macgillivray@intel.com>
Signed-off-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Daniel Martin Buckley <daniel.m.buckley@intel.com>
|
|
|
|
The problem is that we share the same ProxResourceHelper
for both VNF and TG.
For VNF we want to talk to resource.py and get collectd KPIs.
For TG we need to read from the queue the TG calculated KPIs and
we also want collectd KPIs.
workaround is to use a different method name collect_collectd_kpi
for VNFs
Change-Id: Icc2132758e37ce210f5600a0cd433077930208e5
Signed-off-by: Ross Brattain <ross.b.brattain@intel.com>
|
|
the prox files were being found correctly.
if we use find_relative_file they will lookup
relative to the task_path
Change-Id: Ifde5d07df5ccfbfeba015b2f43bd8b53e89a00b7
Signed-off-by: Ross Brattain <ross.b.brattain@intel.com>
|
|
we generate the prox_config_dict in the _run Process,
but we also need it in the _traffic_runner Process to
get core info.
use a queue to pass the config list between the processes
enable collect_kpi
Change-Id: Ibaf41d606e559a87addf43d6ddaed206dbd2d20c
Signed-off-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Edward MacGillivray <edward.s.macgillivray@intel.com>
|
|
Add a new PortPair class to resolve the
topology into list of public and private ports.
Before we were calculating public/private in multiple
locations and using different conventions.
In addition for all the DPDK test we need to use the DPDK
port number and no rely on interface ordering or interface naming
conventions.
We used to use xe0 -> 0, xe1 -> 1, etc. This is not the DPDK port
number.
Use the new dpdknicbind_helper class to parse the output of
dpdk-devbind.py to find the actual DPDK port number at runtime.
We then use this DPDK port number to correctly calculate the
port_mask_hex.
The port mask maps the DPDK port num (PMD ID) to the LINK ID
used in the pipeline config
We also need to make sure we only use the interfaces matched to the
topology and not use all the interfaces, because in some cases we will
have unused interfaces. In particular TRex always requires an even
number of interfaces, so for single port TRex tests we have to create
the second port and not use it.
Thus we had to modify the traffic generator stats code to only dump
stats for used ports and no unused ports.
Ixia was using interface ordering to map to Ixia ports, instead we use
the dpdk_port_num which must be hardcoded for Ixia.
Renamed traffic_profile.execute to traffic_profile.execute_traffic so
we can trace the code easier.
We pass the port used by the traffic profile to generate_samples so we
don't get stats for unused ports.
Fixed up vPE config creation and bring up issues.
Fixed up CGNAPT and UDP_Replay to work correctly.
Tested with 4-port scale-out
Change-Id: I2e4f328bff2904108081e92a4bf712333fa73869
Signed-off-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Edward MacGillivray <edward.s.macgillivray@intel.com>
|
|
JIRA: YARDSTICK-802
Addition of Prox L2Fwd, MPLS test cases for BM and Heat.
updates:
Most of tg_prox and prox_vnf were absorbed into the base classes.
delete most of ProxDpdkVnfSetupEnvHelper, it is handled by DpdkVnfSetupEnvHelper baseclass
use standard _build_pipeline_kwargs methods
don't use terminate() use baseclass version
add new method kill_vnf that runs pkill -x
replace resource_helper.execute() with vnf_execture for dumping stats
In order to share code between tg_prox and vnf_prox
refactor to have tg_prox hold and wrap a ProxApproxVnf instance and call
methods on that class. Do this instead of multiple-inheritance.
Implement ProxApproxVnf.terminate() using prox socket command
based exit, (stop_all, quit, force_quit).
vnf_execute calls resource_helper.execute() which calls
socket methods on the sut object.
Since tg_prox wraps the VNF object, we can call
terminate on the VNF object and it should work correctly.
move prox config generation to parent process
we need to get core number info from config file
inside the TG processes, so we need to generate
the config in the parent process so the data is
copied to the child during the fork.
moved more config file methods to the setup_helper class.
we run force_quit after quit, so the socket should already be closed
this will trigger socket error, so add _ignore_errors option for
vnf_execute to ignore socket errors
Fixed the terminate issue. Added MPLS tests.
Added TG Stats in_packet/out_packet
Fixed compile (pep8) issues
Fixed MPLS TG port stats, in/out packets
Added Grafana dashboards for L2FWD and MPLS
Traffic profiles modified for tolerated loss and
precision as per DATS tests.
Added unit test case for Mpls
Single port test stats collection support.
Change-Id: Idd9493f597c668a3bb7d90e167e6a418546106e8
Signed-off-by: Abhijit Sinha <abhijit.sinha@intel.com>
Signed-off-by: Ross Brattain <ross.b.brattain@intel.com>
|
|
PROX was added to samplevnf project
https://git.opnfv.org/samplevnf/tree/VNFs/DPPD-PROX
JIRA: YARDSTICK-638
Change-Id: If9875b1130c6bed87deb8720b0d8b28ede9289d9
Signed-off-by: Ross Brattain <ross.b.brattain@intel.com>
|