2019-03-22 08:13:12 UTC - Barry: Hi, How does Invoker determine how many
containers it can launch?
I found that the number of containers is always maintained at a fixed value,
exceeded will be removed.
Is it related to these configurations? And how to configure it ?
CONFIG_whisk_loadbalancer_invokerUserMemory=1024m
CONFIG_whisk_containerPool_userMemory=1024m
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553242392352900
----
2019-03-22 08:32:24 UTC - Dominic Kim: You can configure this value in
`ansible/environments/distributed/group_vars/all`
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553243544353300
----
2019-03-22 08:32:51 UTC - Dominic Kim: `invoker_user_memory`
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553243571353500
----
2019-03-22 08:42:16 UTC - Barry: What is the relation between
invoker_user_memory size and container number ? e.g. invoker_user_memory = 1024m
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553244136354300
----
2019-03-22 08:42:48 UTC - Dominic Kim: It is calculated based on the memory
limit configured in the action.
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553244168354700
----
2019-03-22 08:43:08 UTC - Dominic Kim: For example, if you utilize 256MB
actions, one invoker can lunch 4 actions.
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553244188355200
----
2019-03-22 08:43:21 UTC - Dominic Kim: If you use 512MB actions, it will lunch
at most 2 actions.
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553244201355600
----
2019-03-22 08:43:56 UTC - Barry: I got it, thanks :grin:
white_check_mark : Dominic Kim
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553244236356100
----
2019-03-22 12:40:42 UTC - Vincent Hou: <!here> I have made the Jenkins pipeline
automatically kicked-off by the pull requests for openwhisk:
<https://builds.apache.org/view/O/view/OpenWhisk/job/OpenWhisk-Playground/>
raised_hands : Carlos Santana
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553258442356900?thread_ts=1553258442.356900&cid=C3TPCAQG1
----
2019-03-22 12:41:29 UTC - Saul Stockhus: well, I'm glad
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553258489357600
----
2019-03-22 12:42:13 UTC - Dave Grove: Do you know what went wrong here?
<https://github.com/apache/incubator-openwhisk/pull/4384> It looks like the
Jenkins job didn’t go, and that will block merging the PR
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553258533358100?thread_ts=1553258442.356900&cid=C3TPCAQG1
----
2019-03-22 12:42:37 UTC - Vincent Hou: However, I thinking of utilizing three
VMs to provide the edge service instead of one VM.
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553258557358700
----
2019-03-22 12:43:01 UTC - Vincent Hou: so that we can have 3 instances running
instead of only one instance running at a time.
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553258581359200?thread_ts=1553258581.359200&cid=C3TPCAQG1
----
2019-03-22 12:44:14 UTC - Vincent Hou: The build is fine. It only failed the
test case I have mentioned about the pipeline.
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553258654359300?thread_ts=1553258442.356900&cid=C3TPCAQG1
----
2019-03-22 12:44:30 UTC - Vincent Hou: Jenkins build does not block any merge.
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553258670359500?thread_ts=1553258442.356900&cid=C3TPCAQG1
----
2019-03-22 12:44:51 UTC - Vincent Hou: Travis is required.
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553258691359700?thread_ts=1553258442.356900&cid=C3TPCAQG1
----
2019-03-22 12:45:11 UTC - Dave Grove: thanks. i misread the reason the merge
was blocked. not enough :coffee: yet
coffeeparrot : Carlos Santana
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553258711359900?thread_ts=1553258442.356900&cid=C3TPCAQG1
----
2019-03-22 12:51:18 UTC - Vincent Hou: np
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553259078360300?thread_ts=1553258442.356900&cid=C3TPCAQG1
----
2019-03-22 13:01:55 UTC - Carlos Santana: Hmm don’t get it. Whole purpose of
the Jenkins PG test is to have a deployment of multiple VMs, make sure we can
deploy. And also add the tests that we can shutdown one iniver and make sure
that actions can continue to run with the other invokers that are still
available
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553259715363100?thread_ts=1553258581.359200&cid=C3TPCAQG1
----
2019-03-22 14:48:48 UTC - chetanm: @Vincent Hou Any guidance on how to validate
changes in Jenkins file
<https://github.com/apache/incubator-openwhisk/pull/4386>
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553266128363700
----
2019-03-22 14:50:47 UTC - Vincent Hou: I personally created other pipeline,
able to accept parameters, and manually launch it to test.
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553266247364700?thread_ts=1553266247.364700&cid=C3TPCAQG1
----
2019-03-22 14:54:01 UTC - chetanm: How can that be done?
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553266441364800?thread_ts=1553266247.364700&cid=C3TPCAQG1
----
2019-03-22 14:56:11 UTC - Vincent Hou: VM1 launches all the services except
invoker, VM2 launches invoker0 and VM3 launches invoker1. At the same time, we
can do VM2 launches all the services except invoker, VM3 launches invoker0 and
VM1 launches invoker1. Furthermore, VM3 launch all the services except
invoker, VM1 launches invoker0 and VM2 launches invoker1. This topology can run
3 instances at the same time without interference.
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553266571365000?thread_ts=1553258581.359200&cid=C3TPCAQG1
----
2019-03-22 14:57:43 UTC - Vincent Hou: First, are you able to create pipeline
for apache jenkins:
<https://cwiki.apache.org/confluence/display/OPENWHISK/How+to+maintain+the+Jenkins+pipeline+for+OpenWhisk>
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553266663365200?thread_ts=1553266247.364700&cid=C3TPCAQG1
----
2019-03-22 14:58:47 UTC - Vincent Hou: @Carlos Santana How can we intentionally
kill one invoker on one VM, what is the command?
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553266727365900
----
2019-03-22 14:59:47 UTC - chetanm: Let me do that first then
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553266787366000?thread_ts=1553266247.364700&cid=C3TPCAQG1
----
2019-03-22 15:01:34 UTC - Vincent Hou: I persoanlly had one for test before
committing:
<https://builds.apache.org/view/O/view/OpenWhisk/job/OpenWhisk-Test/>
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553266894366200?thread_ts=1553266247.364700&cid=C3TPCAQG1
----
2019-03-22 15:02:09 UTC - chetanm: okie would use similar approach then
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553266929366400?thread_ts=1553266247.364700&cid=C3TPCAQG1
----
2019-03-22 15:04:04 UTC - Vincent Hou: I am thinking of the following topology
to orchestrate multiple distributed deployment of openwhisk with 3 VMs: VM1
launches all the services except invoker, VM2 launches invoker0 and VM3
launches invoker1. At the same time, we can do VM2 launches all the services
except invoker, VM3 launches invoker0 and VM1 launches invoker1. Furthermore,
VM3 launch all the services except invoker, VM1 launches invoker0 and VM2
launches invoker1. This topology can run 3 instances at the same time without
interference. @Carlos Santana @Rodric Rabbah Does it make sense to you?
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553267044367500
----
2019-03-22 15:05:08 UTC - Vincent Hou: I insist it will work perfectly.
partyparrot : dubee, Saurab Joshi
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553267108368200
----
2019-03-22 15:06:50 UTC - Vincent Hou: <!here> everyone loves it?
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553267210368700?thread_ts=1553267210.368700&cid=C3TPCAQG1
----
2019-03-22 15:07:32 UTC - Markus Thömmes: please stop using AT here, you're
potentially pinging 1186 people in this channel :pray:
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553267252368800?thread_ts=1553267210.368700&cid=C3TPCAQG1
----
2019-03-22 15:59:47 UTC - Saurab Joshi: how long are we allowed to add sleep to
a function? We see a weird error when we make it sleep for 60s vs 10s.
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553270387369700
----
2019-03-22 16:02:32 UTC - Rodric Rabbah: the default timeout on an action is
60s so if you sleep for that long your night might get killed
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553270552370500
----
2019-03-22 16:04:07 UTC - Saurab Joshi: oh okay that makes sense
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553270647370700
----
2019-03-22 16:04:10 UTC - Saurab Joshi: thanks!
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553270650370900
----
2019-03-22 16:07:10 UTC - Rodric Rabbah: Sorry I’m not following. What is the
goal with the deployment?
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553270830371700?thread_ts=1553267210.368700&cid=C3TPCAQG1
----
2019-03-22 17:30:41 UTC - Carlos Santana: It doesn’t buy us anything other than
what Travis provides today
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553275841373300?thread_ts=1553258581.359200&cid=C3TPCAQG1
----
2019-03-22 17:31:05 UTC - Carlos Santana: Travis is already running tests a lot
of them for every PR
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553275865374000?thread_ts=1553258581.359200&cid=C3TPCAQG1
----
2019-03-22 17:32:11 UTC - Carlos Santana: I preferred to use Apache Jenkins for
tests that can’t be done in Travis like performance, long tests longer than
Travis max, throughput distribution among invokers in different VMs
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553275931375900?thread_ts=1553258581.359200&cid=C3TPCAQG1
----
2019-03-22 20:23:38 UTC - Bin Wang: Hi. I’m running self-hosted Openwhisk on
Kubernetes. The documentation says the Triggers limit is 60 per user per
minute:
<https://github.com/apache/incubator-openwhisk/blob/master/docs/reference.md#triggers>
This seems too low for production use.
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553286218376600
----
2019-03-22 20:24:00 UTC - Bin Wang: Is there a way to increase this limit? I
tried things like `kubectl -n openwhisk -ti exec owdev-wskadmin -- wskadmin
limits set userA --firesPerMinute 10000` but it didn’t work
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553286240376800
----
2019-03-22 20:27:24 UTC - Dave Grove: You can set different limits when you
deploy the Helm chart. Provide different values in your `mycluster.yaml` for
some or all of the values at
<https://github.com/apache/incubator-openwhisk-deploy-kube/blob/master/helm/openwhisk/values.yaml#L55-#L79>
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553286444378000
----
2019-03-22 20:31:50 UTC - Bin Wang: @Dave Grove Thanks! Maybe we can add this
info to the System limits section in the documentation?
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553286710379100
----
2019-03-22 20:41:43 UTC - Dave Grove: Should probably add a section to
<https://github.com/apache/incubator-openwhisk-deploy-kube/blob/master/docs/configurationChoices.md>.
+1 : Carlos Santana
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1553287303379500
----
