Labelling this PR as size/M

GCE e2e build/test passed for commit 622a28c.

• Build Log
• Test Artifacts
• Internal Jenkins Results

FWIW cadvisor flags are deprecated #19432

I think this changes work for me. Here is how I tried it:

1. rebuild a km docker image with the change in it, plus setting these two flags when launching the scheduler: --housekeeping_interval=50s --global_housekeeping_interval=100s
2. deploy mesos, and kubernetes as a framework on mesos using the km image. docker exec into the km container:

• check the km scheduler process:
  00:00:00 /opt/kubernetes/km scheduler --housekeeping_interval=50s --global_housekeeping_interval=100s --address=10.143.197.73 --mesos-master=10.143.197.73:5050 --etcd-servers=http://10.143.197.73:4001 --mesos-user=root --api-servers=10.143.197.73:8888 --cluster-dns=10.10.10.10 --cluster-domain=cluster.local --v=4
• validated in scheduler.log:

I0316 01:50:42.370195 47 service.go:503] prepared executor command "./km" with args '[./km minion --run-proxy=true --proxy-
bindall=false --proxy-logv=0 --proxy-mode=userspace --path-override= --max-log-size=10Mi --max-log-backups=5 --max-log-age=7 --c
onntrack-max=0 --conntrack-tcp-timeout-established=0 --api-servers=10.143.197.73:8888 --v=0 --allow-privileged=false --suicide-t
imeout=20m0s --mesos-launch-grace-period=5m0s --mesos-cgroup-prefix=mesos --cadvisor-port=4194 --sync-frequency=10s --contain-po
d-resources=true --enable-debugging-handlers=true --cluster-dns=10.10.10.10 --cluster-domain=cluster.local --housekeeping_interv
al=50s --global_housekeeping_interval=1m40s]'

I'm not sure how to check minion/server.go log, else, I can probably validate this line too:

log.V(2).Infof("Spawning hyperkube %v with args '%+v'", server, args)

you can look at the stderr of the mesos sandbox that's running the
executor, it shows the command lines executed for the kubelet and kube-proxy

On Tue, Mar 15, 2016 at 10:23 PM, Lin Sun notifications@github.com wrote:

I think this changes work for me. Here is how I tried it:

rebuild a km docker image with the change in it, plus setting these
two flags when launching the scheduler: --housekeeping_interval=50s
--global_housekeeping_interval=100s
2.

deploy mesos, and kubernetes as a framework on mesos using the km
image. docker exec into the km container:

check the km scheduler process:
00:00:00 /opt/kubernetes/km scheduler --housekeeping_interval=50s
--global_housekeeping_interval=100s --address=10.143.197.73
--mesos-master=10.143.197.73:5050 --etcd-servers=
http://10.143.197.73:4001 --mesos-user=root --api-servers=
10.143.197.73:8888 --cluster-dns=10.10.10.10
--cluster-domain=cluster.local --v=4

validated in scheduler.log:

I0316 01:50:42.370195 47 service.go:503] prepared executor command "./km"
with args '[./km minion --run-proxy=true --proxy-
bindall=false --proxy-logv=0 --proxy-mode=userspace --path-override=
--max-log-size=10Mi --max-log-backups=5 --max-log-age=7 --c
onntrack-max=0 --conntrack-tcp-timeout-established=0 --api-servers=
10.143.197.73:8888 --v=0 --allow-privileged=false --suicide-t
imeout=20m0s --mesos-launch-grace-period=5m0s --mesos-cgroup-prefix=mesos
--cadvisor-port=4194 --sync-frequency=10s --contain-po
d-resources=true --enable-debugging-handlers=true
--cluster-dns=10.10.10.10 --cluster-domain=cluster.local
--housekeeping_interv al=50s --global_housekeeping_interval=1m40s]'

I'm not sure how to check minion/server.go log, else, I can probably
validate this line too:

log.V(2).Infof("Spawning hyperkube %v with args '%+v'", server, args)

—
You are receiving this because you were assigned.
Reply to this email directly or view it on GitHub
#22974 (comment)

@jdef, I'm hitting an environmental issue where the executor on the host times out when downloading the km binary... seems the timeout is 1 min. trying to see if i can increase the network speed or tune the 1 min timeout value.

try the executor_registration_timeout slave option?

On Wed, Mar 16, 2016 at 11:05 AM, Lin Sun notifications@github.com wrote:

@jdef https://github.com/jdef, I'm hitting an environmental issue where
the executor on the host times out when downloading the km binary... seems
the timeout is 1 min. trying to see if i can increase the network speed or
tune the 1 min timeout value.

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#22974 (comment)

Sorry for the delay. I updated the network speed of the env, and the stderr log looks good:

I0317 03:08:12.063347 2850 task.go:220] starting task process "/opt/ibm/mesos/work/slaves/d58a68eb-5105-4be7-9ae8-577d5b24e9b2-S0/frameworks/d58a68eb-5105-4be7-9ae8-577d5b24e9b2-0001/executors/6d493c88c0ac86b3_k8sm-executor/runs/22ac059b-19ff-4ada-bc4d-65db178f3235/km" with args '[executor --api-servers=10.143.100.223:8888 --v=0 --allow-privileged=false --suicide-timeout=20m0s --mesos-launch-grace-period=5m0s --cadvisor-port=4194 --sync-frequency=10s --enable-debugging-handlers=true --cluster-dns=10.10.10.10 --cluster-domain=cluster.local --hostname-override=10.143.100.238 --kubelet-cgroups= --cgroup-root=/mesos/22ac059b-19ff-4ada-bc4d-65db178f3235 --housekeeping_interval=50s --global_housekeeping_interval=1m40s]'
I0317 03:08:12.066579 2850 task.go:257] task started executor

great. i'll mark this as mergeable. we missed the 1.2 cutoff, but this can
probably get cherrypicked for 1.2.1

On Wed, Mar 16, 2016 at 11:12 PM, Lin Sun notifications@github.com wrote:

Sorry for the delay. I updated the network speed of the env, and the
stderr log looks good:

I0317 03:08:12.063347 2850 task.go:220] starting task process
"/opt/ibm/mesos/work/slaves/d58a68eb-5105-4be7-9ae8-577d5b24e9b2-S0/frameworks/d58a68eb-5105-4be7-9ae8-577d5b24e9b2-0001/executors/6d493c88c0ac86b3_k8sm-executor/runs/22ac059b-19ff-4ada-bc4d-65db178f3235/km"
with args '[executor --api-servers=10.143.100.223:8888 --v=0
--allow-privileged=false --suicide-timeout=20m0s
--mesos-launch-grace-period=5m0s --cadvisor-port=4194 --sync-frequency=10s
--enable-debugging-handlers=true --cluster-dns=10.10.10.10
--cluster-domain=cluster.local --hostname-override=10.143.100.238
--kubelet-cgroups=
--cgroup-root=/mesos/22ac059b-19ff-4ada-bc4d-65db178f3235 --housekeeping_interval=50s
--global_housekeeping_interval=1m40s]'
I0317 03:08:12.066579 2850 task.go:257] task started executor

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#22974 (comment)

Automatic merge from submit-queue

@jdef Why would you like this to be cherrypicked into 1.2?

Some folks from IBM are doing some performance work and want the ability to
change these settings. We barely missed the cutoff for 1.2. The changeset
is relatively small and is backwards compat.

What's the criteria for getting things cherrypicked into 1.2?

On Wed, Mar 23, 2016 at 6:23 PM, Brian Grant notifications@github.com
wrote:

@jdef https://github.com/jdef Why would you like this to be
cherrypicked into 1.2?

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#22974 (comment)

Commit 50b41b6 found in the "release-1.2" branch appears to be this PR. Removing the "cherrypick-candidate" label. If this s an error find help to get your PR picked.