Closed jaseemabid closed 5 years ago
@hrishin Are you sure you could access the app? I cant.
This took quite some frustrating effort to dig into. We forked the kubernetes-pipeline-plugin added a lot of expose controller specific code into it. I was looking at the upstream code for a long time :man_shrugging:
@maxandersen I think we broke vertx quickstarts on prod a long time ago and nobody even noticed. I don't know enough about this to take a call on this issue right now.
/ cc: @pradeepto @kbsingh
@jaseemabid so according to you this has been broken since October 2017?
@aslakknutsen Let me try to reproduce this with some other account.
@aslakknutsen I tried the same with @jarifibrahim's account, but we never got past All nodes of label ‘mavenvertx2_2’ are offline
even after several attempts. @rupalibehera Can you please try it out once?
@jaseemabid, I see the same error but, I am able access the app's deployed in stage and run environment.
command: echo | exposecontroller --watch-namespace rbehera-1-run --exposer Route out: Please enter Username: Please enter Password:
command: echo | exposecontroller --watch-namespace rbehera-1-run --exposer Route err: I0301 15:29:35.178807 756 exposecontroller.go:56] Using build: '2.3.27'
I0301 15:29:35.191601 756 config.go:20] No /etc/exposecontroller/config.yml file found. Will try to figure out defaults
I0301 15:29:35.191635 756 exposecontroller.go:125] Running in : `rbehera-1-run`
W0301 15:29:35.222531 756 controller.go:478] Failed to load url https://openshift.default.svc/.well-known/oauth-authorization-server got: Get https://openshift.default.svc/.well-known/oauth-authorization-server: x509: certificate signed by unknown authority
E0301 15:29:35.223477 756 controller.go:186] Failed to list nodes to detect minishift: the server does not allow access to the requested resource (get nodes)
I0301 15:29:35.223491 756 controller.go:118] Using OAuth Authorize URL:
W0301 15:29:35.223497 756 controller.go:120] Please use $OAUTH_AUTHORIZE_URL to define the OAuth Authorize URL!
W0301 15:29:35.224265 756 controller.go:251] Could not find kubernetes service in the default namespace so we could not detect whether to use http or https as the apiserver protocol. Error: the server does not allow access to the requested resource (get services kubernetes)
I0301 15:29:35.224303 756 controller.go:534] starting expose controller
I0301 15:29:40.224491 756 controller.go:542] stopping expose controller```
Am also seeing this same error:
I0302 14:39:44.973538 1033 exposecontroller.go:125] Running in : `ldimaggi-stage`
W0302 14:39:45.006297 1033 controller.go:478] Failed to load url https://openshift.default.svc/.well-known/oauth-authorization-server got: Get https://openshift.default.svc/.well-known/oauth-authorization-server: x509: certificate signed by unknown authority
E0302 14:39:45.007310 1033 controller.go:186] Failed to list nodes to detect minishift: the server does not allow access to the requested resource (get nodes)
I0302 14:39:45.007329 1033 controller.go:118] Using OAuth Authorize URL:
W0302 14:39:45.007335 1033 controller.go:120] Please use $OAUTH_AUTHORIZE_URL to define the OAuth Authorize URL!
W0302 14:39:45.008074 1033 controller.go:251] Could not find kubernetes service in the default namespace so we could not detect whether to use http or https as the apiserver protocol. Error: the server does not allow access to the requested resource (get services kubernetes)
I0302 14:39:45.008110 1033 controller.go:534] starting expose controller
I0302 14:39:50.008333 1033 controller.go:542] stopping expose controller
But - the app is deployed, is reachable, and runs on both stage and run.
yes @jaseemabid http://vertxhttpbooster-hshinde-stage.8a09.starter-us-east-2.openshiftapps.com
Here is the app, like others I'm also facing expose controller error logs but the app is getting deployed in the staging environment.
Now I've to find out why my deployments are failing. How long does it take usually for a deployment to come up @hrishin?
@ldimaggi why is this not a bug
? What type
should it be?
I can deploy a 'Vertx health check' app, but the logs show very similar errors.
I0305 09:21:35.698243 956 config.go:20] No /etc/exposecontroller/config.yml file found. Will try to figure out defaults
I0305 09:21:35.698291 956 exposecontroller.go:125] Running in : `jabid-run`
W0305 09:21:35.731148 956 controller.go:478] Failed to load url https://openshift.default.svc/.well-known/oauth-authorization-server got: Get https://openshift.default.svc/.well-known/oauth-authorization-server: x509: certificate signed by unknown authority
E0305 09:21:35.732167 956 controller.go:186] Failed to list nodes to detect minishift: the server does not allow access to the requested resource (get nodes)
I0305 09:21:35.732211 956 controller.go:118] Using OAuth Authorize URL:
W0305 09:21:35.732217 956 controller.go:120] Please use $OAUTH_AUTHORIZE_URL to define the OAuth Authorize URL!
W0305 09:21:35.732993 956 controller.go:251] Could not find kubernetes service in the default namespace so we could not detect whether to use http or https as the apiserver protocol. Error: the server does not allow access to the requested resource (get services kubernetes)
I0305 09:21:35.733032 956 controller.go:534] starting expose controller
I0305 09:21:40.733226 956 controller.go:542] stopping expose controller
@ldimaggi Out dependency on exposecontroller is most probably a bug and should be fixed.
@jaseemabid here @joshuawilson is pointing you out that we need to add a label on issue for classifications when you log an issue on openshift.io/
Thanks @chmouel that is correct. Now that you made it a type/bug
, what is the severity
?
@joshuawilson This is not blocking anyone from deploying code, but is definitely pointing towards a few other problems. SEV2 or 3. Someone from build team needs to investigate this and I don't when we will be able to.
Is the issue noticeable and annoying but you can work around it? then SEV2 Is the issue noticeable but you can live with it? then SEV3 is the issue not obvious or cosmetic? then SEV4
How long does it take usually for a deployment to come up?
Around 12-14 minutes for the entire build. After that app is immediately getting available to access.
@sthaha I've updated the title and description. This is not a blocker anymore, just something wrong related to deployments.
I took a measurement of how long it takes to execute these statemetns and fail and it seems the saving is only 6 seconds
00:05:52.855 Running: echo | exposecontroller --watch-namespace sunil-thaha-stage --exposer Route
00:05:58.095 command: echo | exposecontroller --watch-namespace sunil-thaha-stage --exposer Route out: Please enter Username: Please enter Password:
00:05:58.096 command: echo | exposecontroller --watch-namespace sunil-thaha-stage --exposer Route err: I0315 09:03:47.926726 619 exposecontroller.go:56] Using build: '2.3.27'
00:05:58.096 I0315 09:03:47.948940 619 config.go:20] No /etc/exposecontroller/config.yml file found. Will try to figure out defaults
00:05:58.096 I0315 09:03:47.948976 619 exposecontroller.go:125] Running in : `sunil-thaha-stage`
00:05:58.096 W0315 09:03:48.030514 619 controller.go:478] Failed to load url https://openshift.default.svc/.well-known/oauth-authorization-server got: Get https://openshift.default.svc/.well-known/oauth-authorization-server: x509: certificate signed by unknown authority
00:05:58.096 E0315 09:03:48.031446 619 controller.go:186] Failed to list nodes to detect minishift: the server does not allow access to the requested resource (get nodes)
00:05:58.096 I0315 09:03:48.031470 619 controller.go:118] Using OAuth Authorize URL:
00:05:58.096 W0315 09:03:48.031476 619 controller.go:120] Please use $OAUTH_AUTHORIZE_URL to define the OAuth Authorize URL!
00:05:58.096 W0315 09:03:48.032236 619 controller.go:251] Could not find kubernetes service in the default namespace so we could not detect whether to use http or https as the apiserver protocol. Error: the server does not allow access to the requested resource (get services kubernetes)
00:05:58.096 I0315 09:03:48.032268 619 controller.go:534] starting expose controller
00:05:58.096 I0315 09:03:53.032450 619 controller.go:542] stopping expose controller
00:05:58.096
00:05:58.096 exposecontroller completed
@pradeepto I am going move this task back to backlog since I am working on the sev-1 : https://github.com/openshiftio/openshift.io/issues/2427
@sthaha To make my point clear, I'm not worried about the performance implications, this is about us doing something obviously wrong with a software we aren't even suppose to use, but somehow the build succeeded because of a retry logic somewhere in the Jenkinsfile.
@jaseemabid I was one worried if there was perf issues from this, we all agree we should still fix it since it's ugly as...
This would go away with OSIO pipeline library. As sooner or later we will move Java projects to new us new pipeline library.
cc: @sthaha @rupalibehera
Going to mark this as "won't fix" since this will be solved in by the new pipeline - https://github.com/fabric8io/osio-pipeline
Expose controller is something we believed we weren't even using in production, but I'm noticing errors related to ec in the deployment step. Relevant part of the logs are
cc: @maxandersen.
@kbsingh I did this immediately after a tenant reset, so this probably affects every user on osio.
Complete logs: