The time we measure is the time to launch and delete the whole
deployment, not the 'pod'. Change the graph titles to be more
accurate.
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
Rather than duplicate the same aes() over all the parts of the ggplot(),
add the aes to the base ggplot, and then default to using that for
all the other additions.
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
geom_smooth can generate a mass of warnings when applied to a small
dataset. Use the same test (>20 samples) we do for 'do we draw dots',
but in the inverse - so, we either draw dots (small set) or a smoothed
overlayed line (bit sets).
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
Now the data has the ability to 'step' (that is, increment
the pod count by >1 between samples), use that n_pod value
for the x-axis, rather than a +1 count, so the x-axis makes
a lot more sense.
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
If a RUNTIME is not set, then save it as 'default' in the JSON
rather than an empty string, as the emptry string does not render
well later in any graphing.
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
Add invididual date entries to each statistics sample, to
suppliment the top level date entry. This helps order the
samples if we are flattening the data into say Elasticsearch.
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
When 'grace' handling was added, it got missed from the parallel
script (which shares the bb.*.in files). Add it in to fix the test.
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
proxy_ip_list variable had scope issues, used as global variable
initially then as a local variable. Changed all to global.
Signed-off-by: root <root@csterret-desk1.amr.corp.intel.com>
This commit replaces our static manifests with Kustomize
versions and adds scripting to apply them.
Signed-off-by: Justin Scott <justin.a.scott@intel.com>
We were storing some JSON data (such as NUM_PODS) as strings.
This would default their type into the Elastic DB as strings,
and not ints. May as well store ints as ints, so make that so.
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
The boot 'points' code was accidentally updating the cpu plot
(copy and paste error). Correct it to update the correct chart.
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
Test how long it takes to launch, and delete, a deployment
running up 'n' pods. Used to ascertain if parallel launch times
are linear with number of pods or not.
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
We were carrying a bunch of legacy docker based functions
that we do not use. Drop them. We can always add them back
in later if we find we do need some.
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
We are very likely to re-use the k8s api code in many/all of the
tests, so move common code out to a common libray file.
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
Capture some of the k8s setup information in the JSON results
files. Add sections for:
- kubectl version
- kubectl get nodes
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
If we have $RUNTIME set in the env, then we set up the pod
RuntimeClass, otherwise, just use the default node runtimes.
Clean up some of the un-used library code around docker/RUNTIME
as well whilst we are there.
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
We need to check when we are launching our first instance of a
deployment when using the k8s API directly, as we have to modify
the URL for the first instance. Rather than check for '1', let's
check for 'STEP', as technically you can run the loop starting at
a non-1 index (particularly useful to run with NUM_PODS==STEP, as
you then get a 'single shot' test of 'n' containers).
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
Add the first basic report generator code, utilising R, Rmarkdown
and pandoc to do the data processing and pdf report generation.
Fixes: #121
Signed-off-by: Graham Whaley <graham.whaley@intel.com>
This setup scripts have been using cri-o all this while as the pod
controller/manager system. Recent past cri-o has had some issues with
kata-deploy (A restart of the service will not be able to re-connect
with the existing pods and restart all of them including kata-deploy,
which will hit an endless loop). Moving to containerd as default for
now.
Firecracker cannot be used with a released version of containerd as
there is no default block based snapshotter available today. If you wish
to use cri-o make sure you set `RUNNER=crio` in your environment prior
to using the script.
Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
Modifying rp_filter setting on host seems to be causing a routing issue
when connecting to a pod from remote node. Until we figure out the best
way to solve the issue, we are configuring canal to ignore default
setting of 2
Fixes: #102
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
ClearLinux now has cri-o version 1.14.1 which allows multiple plugin
locations and also creates /opt/cni/bin by default. We no longer need
the hacks for them
Fixes: #82
Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
Currently net.ipv4.conf.all.rp_filter is set to 2 in Clear and
calico-node fails to come up unless that value is either 0 or 1.
Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
- If stop CRI service if is running
- Restart the CRI service only if enabled
- Do not enable crio on reset_stack.sh this should be only part of the
setup.
Signed-off-by: Jose Carlos Venegas Munoz <jose.carlos.venegas.munoz@intel.com>
kubeadm autodetects the socket path based on defaults from well known
CRI servers.
Signed-off-by: Jose Carlos Venegas Munoz <jose.carlos.venegas.munoz@intel.com>
Graham Whaley