2019-03-14 03:00:14 UTC - Jin Choi: @Rodric Rabbah I am running Openwhisk system on my servers and here's one question for you. For Openwhisk deployment using ansible can we make it BLUE-GREEN deployment? Ansible playbooks in `${OPENWHISK_HOME}/ansible` doesn't seem so. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552532414119600 ---- 2019-03-14 03:01:12 UTC - Rodric Rabbah: I think you’re right there isn’t a playbook for it.
https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552532472120700 ---- 2019-03-14 03:01:36 UTC - Jin Choi: How does IBM deploy Openwhisk without downtime? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552532496121400 ---- 2019-03-14 03:02:37 UTC - Rodric Rabbah: blue green is a good bet :) https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552532557121800 ---- 2019-03-14 03:04:02 UTC - Jin Choi: Once we migrate Openwhisk onto k8s, it will do the whole magic for us, right? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552532642124100 ---- 2019-03-14 03:04:46 UTC - Jin Choi: While we are here stuck with ansible for the time being... I was wondering if I could provide zero downtime deployment experience for my customers. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552532686125200 ---- 2019-03-14 03:05:34 UTC - Rodric Rabbah: you can deploy an outer router (another nginx) and switch the active cluster by reloading the config file https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552532734126300 ---- 2019-03-14 03:06:09 UTC - Jin Choi: Is that how it's done in IBM? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552532769126800 ---- 2019-03-14 03:06:29 UTC - Jin Choi: One charming solution, BTW https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552532789127100 ---- 2019-03-14 03:08:46 UTC - Jin Choi: In that case I need 2 seperate clusters of Openwhisk, and those two need to share the same one-and-only couchdb cluster. Am I getting it right? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552532926128700 ---- 2019-03-14 03:09:08 UTC - Rodric Rabbah: our earliest solutions did just that - I can’t say what ibm does today https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552532948129200 ---- 2019-03-14 03:09:36 UTC - Jin Choi: You mean, it's credential, right? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552532976130300 ---- 2019-03-14 03:09:51 UTC - Rodric Rabbah: I think couch could be shared - you can replicate fully the rest as one option https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552532991131000 ---- 2019-03-14 03:20:40 UTC - Jin Choi: With the suggested solution by @Rodric Rabbah I am worried if I could switch the Openwhisk clusters safely. It is the key making sure every actions that were being invoked on the BLUE cluster are all served and done. Otherwise some action invocation requests would return generated activation IDs and the result might disappear. Only then I could switch the outer router from BLUE to GREEN. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552533640138400 ---- 2019-03-14 03:20:46 UTC - Jin Choi: What do you think? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552533646138600 ---- 2019-03-14 03:22:39 UTC - Rodric Rabbah: If you use the outer router as the api host then once you switch the active cluster the rest of activations in the now inactive cluster will just drain https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552533759139600 ---- 2019-03-14 04:49:41 UTC - Carlos Santana: We use a shared Cloudant SaaS, we use Cloudflare Internet Services SaaS outer router to toggle the blue green, 0 downtime https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552538981141300 ---- 2019-03-14 04:55:29 UTC - Dominic Kim: @Jin Choi FYI, We(Naver) are deploying OpenWhisk without downtime(Not blue-green) in this way: 1. Remove a controller from the nginx 2. Wait until controller does not handle any activations(with API implemented to fetch the number of ongoing activations) 3. Deploy the controller 4. Readd the controller to nginx. thankyou : Jin Choi https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552539329144100 ---- 2019-03-14 04:55:43 UTC - Dominic Kim: Similar logic is applied to invokers as well. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552539343144500 ---- 2019-03-14 05:00:59 UTC - Jin Choi: @Dominic Kim Do you use the ansible playbooks in `OPENWHISK_HOME/ansible` as it is for the deploying strategy you are saying? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552539659146600 ---- 2019-03-14 05:01:23 UTC - Dominic Kim: Yes https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552539683146800 ---- 2019-03-14 05:08:37 UTC - Jin Choi: And you don't modify this file, `OPENWHISK_HOME/ansible/environments/prod/hosts`, during the deployment? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552540117150200 ---- 2019-03-14 05:08:58 UTC - Dominic Kim: It depends. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552540138150500 ---- 2019-03-14 05:09:18 UTC - Dominic Kim: If we don't provision more servers, we don't touch it. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552540158150900 ---- 2019-03-14 05:19:12 UTC - Carlos Santana: <https://rancher.com/blog/2019/announcing-submariner-multi-cluster-kubernetes-networking/> https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552540752154900 ---- 2019-03-14 05:22:51 UTC - Jin Choi: @Dominic Kim Let's say we don't modify the `hosts` file and we have 3 controller servers here. As far as I know, `controller.yaml` playbook runs the ansible script against all the 3 controller servers. How come do you manage the ansible playbook to pick a controller and deploy onto it only, while not touching the other two? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552540971156900?thread_ts=1552540971.156900&cid=C3UDXSFA6 ---- 2019-03-14 05:27:38 UTC - Carlos Santana: ansible-playbook has some flags I think is —limit you can RTFM to find out more https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552541258158400 ---- 2019-03-14 05:29:08 UTC - Carlos Santana: <https://stackoverflow.com/questions/18195142/safely-limiting-ansible-playbooks-to-a-single-machine> https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552541348158600 ---- 2019-03-14 05:32:12 UTC - Carlos Santana: You can always create a new ansible playbook rolling_update.yaml to do all your rolling update maneuvers if ansible is your thing. We don’t use ansible anymore, more heavy on Jenkins and kubectl these days https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552541532161600 ---- 2019-03-14 05:38:02 UTC - Dominic Kim: You can change the `serial` number. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552541882161800?thread_ts=1552540971.156900&cid=C3UDXSFA6 ---- 2019-03-14 05:38:23 UTC - Dominic Kim: It controls the number of target hosts which will be deployed concurrently. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552541903162000?thread_ts=1552540971.156900&cid=C3UDXSFA6 ---- 2019-03-14 05:39:19 UTC - Dominic Kim: <https://github.com/apache/incubator-openwhisk/blob/adf3ba1b2cb76d2b4eb09eff3f360bf14bd65623/ansible/prereq.yml#L8> https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552541959162200?thread_ts=1552540971.156900&cid=C3UDXSFA6 ---- 2019-03-14 05:48:20 UTC - Jin Choi: @Carlos Santana Thanks for your suggestion. You are very kind. +1 : Carlos Santana https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552542500162800 ---- 2019-03-14 05:55:36 UTC - Jin Choi: Is Openwhisk on Bluemix powered on k8s? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552542936164000?thread_ts=1552542936.164000&cid=C3UDXSFA6 ---- 2019-03-14 05:57:11 UTC - Jin Choi: The docs on `incubator-openwhisk-deploy-kube` say lots of components (apigateway, couchdb, kakfa, kakfaprovider, nginx, redis) are not eligible for replicaCount more than 1 yet. It means `not production ready` for high traffic? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552543031165300 ---- 2019-03-14 06:05:21 UTC - Carlos Santana: Yes, our cloud name change from Bluemix to “IBM Cloud” yes we use the normal Kubernetes SaaS IKS “IBM Cloud Kubernetes Service” <https://twitter.com/IKS_IBMCloud> https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552543521167500?thread_ts=1552542936.164000&cid=C3UDXSFA6 ---- 2019-03-14 06:08:59 UTC - Carlos Santana: Most of dependencies we use a SaaS, IBM Cloud API Management based on APIGateway , Cloudant based on couchdb, IBM Cloud Databases for Redis, etc... https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552543739170500?thread_ts=1552542936.164000&cid=C3UDXSFA6 ---- 2019-03-14 06:09:15 UTC - Jin Choi: That makes sense... https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552543755170700?thread_ts=1552542936.164000&cid=C3UDXSFA6 ---- 2019-03-14 06:10:47 UTC - Jin Choi: You should have SaaS's for kafka and zookeeper, right? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552543847170900?thread_ts=1552542936.164000&cid=C3UDXSFA6 ---- 2019-03-14 06:13:19 UTC - Jin Choi: With all those auto-scaling SaaS backed up, no wonder why only invokers are scalable on the project `incubator-openwhisk-deploy-kube` yet. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552543999171200?thread_ts=1552542936.164000&cid=C3UDXSFA6 ---- 2019-03-14 06:14:17 UTC - Jin Choi: The project `incubator-openwhisk-deploy-kube` itself not being sufficient for high traffic doesn't mean the Openwhisk on IBM cloud not being HA, if I got it right. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552544057171400?thread_ts=1552542936.164000&cid=C3UDXSFA6 ---- 2019-03-14 09:11:26 UTC - Michele Sciabarra: It means that the helm chart is not yet ready for a production deployment but it does not mean you cannot deploy it in production as all the components are already deployable as scalable https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552554686173000 ---- 2019-03-14 09:12:41 UTC - Michele Sciabarra: my understanding (I do not work for IBM) is that IBM is using internally their already scalable components for Kafka Couchdb but those things are not yet replicated in the Helm Chart. The helm chart allows using external components for Kafka Couchdb etc +1 : Satwik Kolhe https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552554761174700 ---- 2019-03-14 09:40:24 UTC - Carlos Santana: Yep the real workers are the invokers those are the one that you will have more as you want to support more and more concurrent invocations in the system https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552556424176700?thread_ts=1552542936.164000&cid=C3UDXSFA6 ---- 2019-03-14 09:50:54 UTC - Carlos Santana: @Michele Sciabarra that’s correct we use external SaaS for most of the dependencies for IBM. We call it eating our own dog food :smile: yum : Michele Sciabarra https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552557054177800 ---- 2019-03-14 09:51:39 UTC - Carlos Santana: Happy :pie: Day <https://cloud.google.com/blog/products/compute/calculating-31-4-trillion-digits-of-archimedes-constant-on-google-cloud> https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552557099178500 ---- 2019-03-14 15:53:35 UTC - James Thomas: try to go through the apache release process for the JS SDK…. I sorta wish I still drank. see_no_evil : Rodric Rabbah, Shazron Abdullah, Carlos Santana hear_no_evil : Rodric Rabbah https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552578815179700?thread_ts=1552578815.179700&cid=C3UDXSFA6 ---- 2019-03-14 16:35:05 UTC - Carlos Santana: Welcome to the ASF ! https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552581305180300?thread_ts=1552578815.179700&cid=C3UDXSFA6 ---- 2019-03-14 16:54:16 UTC - James Thomas: props to @Vincent Hou for automating most of this! phew. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552582456180600?thread_ts=1552578815.179700&cid=C3UDXSFA6 ---- 2019-03-14 16:57:48 UTC - James Thomas: <https://github.com/apache/incubator-openwhisk-client-js/pull/154> LET’S GET THIS PARTY STARTED https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552582668180800?thread_ts=1552578815.179700&cid=C3UDXSFA6 ---- 2019-03-14 16:59:20 UTC - Carlos Santana: you can also start the mail thread [DISCUSS] to see anyone has any changes they can't wait until next cut, and leave it open for 72 hours. +1 : James Thomas https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552582760181000?thread_ts=1552578815.179700&cid=C3UDXSFA6 ---- 2019-03-14 17:00:27 UTC - Carlos Santana: if no one objects you can cut what's in master in 72 hours to create the tgz https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552582827181200?thread_ts=1552578815.179700&cid=C3UDXSFA6 ---- 2019-03-14 17:02:02 UTC - chetanm: Technically no need to wait for 72 hrs for discuss https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552582922181500?thread_ts=1552578815.179700&cid=C3UDXSFA6 ---- 2019-03-14 17:02:16 UTC - chetanm: But given weekend may be fine to do on Monday https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552582936181700?thread_ts=1552578815.179700&cid=C3UDXSFA6 ---- 2019-03-14 17:10:22 UTC - James Thomas: next step is to write an openwhisk action to run all the verification steps for a release candidate so we don’t have to do it manually. whisking : Carlos Santana openwhisk : Carlos Santana raised_hands : Carlos Santana https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552583422182400 ---- 2019-03-14 17:13:02 UTC - Rodric Rabbah: lol @Dave Grove opened an issue for that, and im on the hook and im so far behind joy : James Thomas https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552583582182900 ---- 2019-03-14 17:14:58 UTC - Carlos Santana: @Rodric Rabbah I think you can close the runtimes [DISCUSS] thread and move on with the next step on packaging and signing the tgz, when is your ETA? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552583698183700 ---- 2019-03-14 17:15:40 UTC - Rodric Rabbah: last monday https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552583740184000 ---- 2019-03-14 17:16:23 UTC - Carlos Santana: already saw the 3 back to the future movies this weekend with my daughter https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552583783184600 ---- 2019-03-14 17:16:46 UTC - Rodric Rabbah: wow Carlos https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552583806185100 ---- 2019-03-14 17:16:50 UTC - Carlos Santana: Awesome that she is starting to like these retro movies crystal_ball : Rodric Rabbah https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552583810185400 ---- 2019-03-14 17:16:54 UTC - Rodric Rabbah: watch 2 less moves and cut some releases! https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552583814185600 ---- 2019-03-14 17:17:06 UTC - Rodric Rabbah: lol https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552583826185900 ---- 2019-03-14 17:19:32 UTC - Carlos Santana: I will signed up once we have everything automated that only takes me a `/` command in Slack to drive a release !, hopefully :robot_face: build with serverless https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552583972187200 ---- 2019-03-14 17:21:06 UTC - Carlos Santana: I thought you need it to give some time to folks to see the email, if your going to open a thread to close in an hour what's the point? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552584066187500?thread_ts=1552578815.179700&cid=C3UDXSFA6 ---- 2019-03-14 17:21:28 UTC - Carlos Santana: all people don't leave in the same time zone https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552584088187800?thread_ts=1552578815.179700&cid=C3UDXSFA6 ---- 2019-03-14 17:36:31 UTC - Dave Grove: What do you think should be done about the client-go/cli blocker for the release. If it was an IBMer, I’d just revert the problem commit to the client-go library and move on. Since its a community commit, I’d prefer to not be unfriendly like that https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552584991189800?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 17:41:52 UTC - Carlos Santana: Have you created an issue, ping the contributor, and gave time to respond and address problem? if that's have being exausted then yes revert it and add a comment to the issue so the reference to the code is not lost and easy to be found if we revisit to make the change https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552585312191600 ---- 2019-03-14 17:41:55 UTC - Rodric Rabbah: is this the activation poll issue? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552585315191800 ---- 2019-03-14 17:43:25 UTC - Carlos Santana: I understand that the change might be good, and the problem lies in another place like integration or test, but do far you gave it a hard try to figured it out, unless if someone else wants to give it a try, you can reach others on CLI land via ML and Slack I guess and cc on the issue https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552585405193200 ---- 2019-03-14 17:48:18 UTC - Rodric Rabbah: im taking a look now https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552585698193900?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 17:49:33 UTC - Dave Grove: sigh. yes. all done already. <https://github.com/apache/incubator-openwhisk-client-go/issues/118> https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552585773194600 ---- 2019-03-14 17:51:43 UTC - chetanm: Ack but discuss thread may just be given lesser time ... For vote you need full 72 hours https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552585903194800?thread_ts=1552578815.179700&cid=C3UDXSFA6 ---- 2019-03-14 17:55:48 UTC - Carlos Santana: I have wasted too much time of previous open projects bikeshed over what "lesser time" means 1 hours, 12 hours, 24 hours, etc.. Dealing with responses with no everyone checks their OSS mailing list daily, etc... Is better not to improvise and set the expectation clear on what's the closing date if there are no objections, then no one get's offended by being close too early. Anyway I'm ok if the person actually doing the real work as release manager says how long he wants to wait and own judgement https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552586148195000?thread_ts=1552578815.179700&cid=C3UDXSFA6 ---- 2019-03-14 18:28:42 UTC - Rodric Rabbah: ugh both tests pass locally https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588122195800?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:30:05 UTC - Rodric Rabbah: when i run ./gradlew compile -PnativeCompile on pr 423 i get a dirty git, with the vendor hashes having changed https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588205196200?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:31:25 UTC - Rodric Rabbah: but if i build with go, it’s ok https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588285196400?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:31:50 UTC - Rodric Rabbah: i get two tests failing though not just one… trying to understand the disparity https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588310196600?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:32:15 UTC - Dave Grove: in travisci it was failing two tests (both of the ConsoleTests). https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588335196800?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:32:39 UTC - Rodric Rabbah: ah https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588359197000?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:32:41 UTC - Rodric Rabbah: ``` penguin [master] openwhisk> travis 505208192 system.basic.WskCliConsoleTests > Wsk Activation Console should show an activation log message for hello world FAILED system.basic.WskCliConsoleTests > Wsk Activation Console should show repeated activations FAILED Done. Your build exited with 1. ``` https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588361197200?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:32:42 UTC - Dave Grove: one test it wanted 1 activation and got 2. second test was wanting 4 activations and getting 5. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588362197400?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:32:55 UTC - Rodric Rabbah: ok so i have reproduces the travis failure https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588375197600?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:34:01 UTC - Rodric Rabbah: the gradlew build for wsk removes the vendor file https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588441197800?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:34:25 UTC - Dave Grove: nice. was there a trick to running the test cases locally using a modified cli? I never figured out how to do that. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588465198000?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:34:56 UTC - Dave Grove: yeah, we seem to be confused about how we are supposed to build a go project and track the dependencies. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588496198200?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:35:02 UTC - Rodric Rabbah: eh - it’s a bit of a :poop: fest… 1. ./gradlew install in openwhisk directory https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588502198400?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:35:11 UTC - Rodric Rabbah: 2. rebuild the cli tests https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588511198600?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:35:36 UTC - Rodric Rabbah: 3. in the test, you can specify the location of the binary to use eg new Wsk(‘new path’) https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588536198800?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:37:40 UTC - Rodric Rabbah: this is what i did https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588660199000?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:37:42 UTC - Rodric Rabbah: ` override val wsk = new Wsk("/Users/rabbah/projects/incubator-openwhisk-cli/wsk")` https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588662199200?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:38:46 UTC - Rodric Rabbah: ok i found the problem https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588726199400?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:38:49 UTC - Rodric Rabbah: :smile: https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588729199600?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:39:50 UTC - Dave Grove: easy fix? It smelled like something silly if we could just reproduce outside of travis, but I never managed to get that to work https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588790199800?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:40:16 UTC - Rodric Rabbah: i posted a hint in the issue https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588816200100?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:40:42 UTC - Dave Grove: trailing newline? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588842200300?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:40:46 UTC - Rodric Rabbah: :smile: https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588846200500?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:41:14 UTC - Rodric Rabbah: lol now i dont know how to fix it locally - how does go work if i locally modify the checkout? do i bump my vendor tag and it’ll be ok? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588874200700?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:41:54 UTC - Dave Grove: if you checked all this stuff into your gopath, it probably just magically works (maybe). https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588914200900?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:42:02 UTC - Rodric Rabbah: ok trying https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588922201100?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:42:40 UTC - Dave Grove: otherwise there’s some black magic that is needed :shrug: https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552588960201300?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:45:05 UTC - Rodric Rabbah: its not obvious to me where the extra `\n` is coming from looking at the PR https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552589105201500?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:45:52 UTC - Rodric Rabbah: oy this might have been a false lead https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552589152201700?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:51:18 UTC - Dave Grove: I have no idea what the cli actually does, but ToSummaryRowString changed from returning a single string to a single string that is : separated and the comment says it is used in wsk activation list. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552589478201900?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:51:23 UTC - Rodric Rabbah: Datetime Activation ID Kind Start Duration Status Entity https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552589483202100?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:51:26 UTC - Rodric Rabbah: i think this is the extra line https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552589486202300?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:51:32 UTC - Dave Grove: the header. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552589492202500?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:52:19 UTC - Rodric Rabbah: trying to fix now https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552589539202700?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:52:43 UTC - Dave Grove: before it used to just say “activations” which maybe got filtered out by the test case https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552589563202900?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:54:06 UTC - Rodric Rabbah: arguably the bug is actually in the openwhisk test harness https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552589646203100?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:54:21 UTC - Rodric Rabbah: which is scaring me a little now https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552589661203300?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:57:21 UTC - Rodric Rabbah: the fix has to be in openwhisk repo https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552589841203500?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:58:05 UTC - Rodric Rabbah: with the old cli, the activation list parser grabs the activation id correctly https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552589885203800?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:58:12 UTC - Rodric Rabbah: but with the new cli, you get this `ArraySeq(Datetime, 2019-03-14)` https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552589892204000?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 18:59:53 UTC - Rodric Rabbah: <https://github.com/apache/incubator-openwhisk/blob/4a8564b9de9fa869a6c22e036b337f9cd2f46221/tests/src/test/scala/common/WskCliOperations.scala#L519-L528> https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552589993204200?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 19:01:09 UTC - Rodric Rabbah: hmm https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552590069204400?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 19:01:33 UTC - Rodric Rabbah: should i make this resilient or just adapt to the new format? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552590093204600?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 19:06:22 UTC - Rodric Rabbah: patched but now we have a phase ordering problem? https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552590382204800?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 19:07:30 UTC - Rodric Rabbah: i guess ill just if/else it and then we can remove the old code https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552590450205000?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 19:20:41 UTC - Dave Grove: messy. arguably when we list activations in “testing”/batch mode we should have an option to suppress the header entirely. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552591241205200?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 20:12:17 UTC - Lars Andersson: @Dave Grove Good thing you didn't follow your instincts. I would never have been able to fix that! Still just trying to teach myself Scala, and on top of that to understand how your test code is organized and to get it to work locally like @Rodric Rabbah did would have been an overwhelming task. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552594337205500?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 20:13:56 UTC - Dave Grove: tell me about it….I couldn’t get the tests to run locally either, which is why I kept doing the binary search via TravisCI PRs https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552594436205700?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 20:15:50 UTC - Rodric Rabbah: @Lars Andersson issues like this really make me regret that we blew up one mono-repo https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552594550205900?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 20:19:33 UTC - Lars Andersson: Yeah, and as I said, I'm still learning Scala and went over the code as good as I could but never found any use of the changed code from doing a poll, which I believed was used to verify the result. Anyway, I will have a look at the PR @Rodric Rabbah created about the format, and I will also go over all the document pages and update to the new format of the list command as well. https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552594773206100?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 20:20:09 UTC - Rodric Rabbah: thanks Lars! https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552594809206300?thread_ts=1552584991.189800&cid=C3UDXSFA6 ---- 2019-03-14 20:20:42 UTC - Lars Andersson: No problem! https://openwhisk-team.slack.com/archives/C3UDXSFA6/p1552594842206500?thread_ts=1552584991.189800&cid=C3UDXSFA6 ----