Closed BenHall closed 6 years ago
Thanks for the report! Looking into it now.
@BenHall a few questions:
Thanks!
This is related to Weave Cloud, using the latest release.
I applied a RBAC fix from the Cortex issue (https://github.com/weaveworks/cortex/issues/392#issuecomment-292713631). This allowed images to be viewed via list-images
but rolling out changes would indicate 0 services required updating. This happened via the CLI and the web UI.
Hrm, I could not replicate the '0 services required updating' issue - instead I get this:
$ fluxctl-0.2.0 --token=<REDACTED> list-images --service=sock-shop/front-end
SERVICE CONTAINER IMAGE CREATED
sock-shop/front-end front-end index.docker.io/weaveworksdemos/front-end
| master-ac9ca707 07 Apr 17 13:06 UTC
| master-3fad5eda 30 Mar 17 08:51 UTC
| master-bdc6f3ff 29 Mar 17 10:45 UTC
'-> 0.3.12 21 Mar 17 11:32 UTC
latest 21 Mar 17 11:32 UTC
master-b2308a3e 21 Mar 17 11:24 UTC
0.3.11 20 Mar 17 12:13 UTC
master-fe7f9828 17 Mar 17 14:24 UTC
0.3.10 16 Mar 17 13:31 UTC
master-30f06738 16 Mar 17 13:30 UTC
$ fluxctl-0.2.0 --token=<REDACTED> release --service=sock-shop/front-end --update-all-images
Submitting release job...
Release job submitted, ID f7636083-65ff-3b4c-4b0c-958ce1eb1778
Status: Failed: sock-shop/front-end: applying definition to sock-shop/front-end: running kubectl: error: group map[federation:0xc8203caf50 :0xc8203cafc0 apps:0xc8203cb030 autoscaling:0xc8203cb110 componentconfig:0xc8203cb1f0 policy:0xc8203cb2d0 rbac.authorization.k8s.io:0xc8203cb340 authorization.k8s.io:0xc8203cb0a0 batch:0xc8203cb180 extensions:0xc8203cb260 authentication.k8s.io:0xc8203cb420] is already registered
Here's as far as we got:
1) Queued.
2) Calculating updates for release.
3) Cloning git repository.
4) Finding defined services.
5) Found service sock-shop/front-end
6) Found 1 services.
7) Looking up images.
8) Will update sock-shop/front-end container front-end: weaveworksdemos/front-end:0.3.12 -> master-ac9ca707
9) Pushing changes.
10) Applying changes.
11) Sending notifications.
12) Failed: sock-shop/front-end: applying definition to sock-shop/front-end: running kubectl: error: group map[federation:0xc8203caf50 :0xc8203cafc0 apps:0xc8203cb030 autoscaling:0xc8203cb110 componentconfig:0xc8203cb1f0 policy:0xc8203cb2d0 rbac.authorization.k8s.io:0xc8203cb340 authorization.k8s.io:0xc8203cb0a0 batch:0xc8203cb180 extensions:0xc8203cb260 authentication.k8s.io:0xc8203cb420] is already registered
Took 19.718682s
== Error ==
Internal error: sock-shop/front-end: applying definition to sock-shop/front-end: running kubectl: error: group map[federation:0xc8203caf50 :0xc8203cafc0 apps:0xc8203cb030 autoscaling:0xc8203cb110 componentconfig:0xc8203cb1f0 policy:0xc8203cb2d0 rbac.authorization.k8s.io:0xc8203cb340 authorization.k8s.io:0xc8203cb0a0 batch:0xc8203cb180 extensions:0xc8203cb260 authentication.k8s.io:0xc8203cb420] is already registered
We don't have a specific help message for the error above.
It would help us remedy this if you log an issue at
https://github.com/weaveworks/flux/issues
saying what you were doing when you saw this, and quoting the message
at the top.
This is with flux 0.2.0 installed on a 1.6.0 k8s cluster via the launch generator:
kubectl apply -f 'https://cloud.weave.works/k8s/flux.yaml?service-token=<REDACTED>'
which includes the required service account.
ISTM that the 'group already registered' error is due to an incompatibility with the kubectl shipped inside fluxd (v1.3.5) - I'm about to open a PR that updates that, and with which I have successfully released sock-shop/front-end to a 1.6 cluster...
The service 0 was from the Flux / Weave Cloud UI
root@master:~/weave-front-end# fluxctl release --service=sock-shop/front-end --update-all-images
Submitting release job...
Release job submitted, ID 15dc7bd0-2ea5-2c2f-9621-2cf1192c33a7
Status: Complete.
Here's what happened:
SERVICE STATUS UPDATES
Took 18.771451s
root@master:~/weave-front-end# fluxctl list-images --service=sock-shop/front-end
SERVICE CONTAINER IMAGE CREA
TED
sock-shop/front-end front-end 2886795273-4567-kitek01.environments.katacoda.com/root/front-end
| 1d8e9cd73cc781c91aa4fcbc87afda2a464da343 10 A
pr 17 16:50 UTC
'-> latest 28 O
ct 16 20:22 UTC
root@master:~/weave-front-end#
Dry Run: Calculating plan to release sock-shop/front-end
Release job submitted, ID 855b5454-0df1-abdd-f381-9c10287f4b2b
Status: Complete.
1) Queued.
2) Calculating updates for release.
3) Cloning git repository.
4) Finding defined services.
5) Found 0 services.
6) Looking up images.
7) No updates to do, finishing.
Success
Releasing sock-shop/front-end
Release job submitted, ID 000891f0-0c74-e6df-477e-d18665a180fc
Status: Complete.
1) Queued.
2) Calculating updates for release.
3) Cloning git repository.
4) Finding defined services.
5) Found 0 services.
6) Looking up images.
7) No updates to do, finishing.
Success
Sorry, didn't mean to close this yet - you clearly have some other issue as well (although you would hit the k8s 1.6 problem once we get past whatever it is)
@BenHall can you paste the result of running kubectl --namespace sock-shop get all -o wide
against that cluster? Thanks
NAME READY STATUS RESTARTS AGE IP NODE
po/carts-153328538-fvncf 1/1 Running 0 6m 10.40.0.2 node
po/carts-db-1968389027-wlpvg 1/1 Running 0 6m 10.40.0.1 node
po/catalogue-114596073-w6h21 1/1 Running 0 6m 10.40.0.7 node
po/catalogue-db-1956862931-l4ktv 1/1 Running 0 6m 10.40.0.4 node
po/front-end-537982664-s53gw 1/1 Running 0 5m 10.40.0.3 node
po/orders-2365168879-mx6b8 1/1 Running 0 5m 10.40.0.6 node
po/orders-db-3036494983-88vc9 1/1 Running 0 5m 10.40.0.5 node
po/payment-1968871107-rfgpn 1/1 Running 0 5m 10.40.0.8 node
po/queue-master-2798459664-bfz41 0/1 ContainerCreating 0 5m <none> node
po/rabbitmq-2262854187-ghnld 0/1 ContainerCreating 0 5m <none> node
po/shipping-2899287913-51t6g 0/1 ContainerCreating 0 5m <none> node
po/user-468431046-ff8sx 0/1 ContainerCreating 0 5m <none> node
po/user-db-1166754267-t12k1 0/1 ContainerCreating 0 5m <none> node
NAME CLUSTER-IP EXTERNAL-IP PORT(S) AGE SELECTOR
svc/carts 10.107.86.207 <none> 80/TCP 6m name=carts
svc/carts-db 10.103.227.2 <none> 27017/TCP 6m name=carts-db
svc/catalogue 10.108.231.190 <none> 80/TCP 6m name=catalogue
svc/catalogue-db 10.101.213.150 <none> 3306/TCP 6m name=catalogue-db
svc/front-end 10.107.188.158 ,172.17.0.13 80:30284/TCP 5m name=front-end
svc/orders 10.111.22.215 <none> 80/TCP 5m name=orders
svc/orders-db 10.100.28.123 <none> 27017/TCP 5m name=orders-db
svc/payment 10.104.246.205 <none> 80/TCP 5m name=payment
svc/queue-master 10.97.136.240 <none> 80/TCP 5m name=queue-master
svc/rabbitmq 10.111.195.90 <none> 5672/TCP 5m name=rabbitmq
svc/shipping 10.107.14.149 <none> 80/TCP 5m name=shipping
svc/user 10.103.100.168 <none> 80/TCP 5m name=user
svc/user-db 10.106.209.225 <none> 27017/TCP 5m name=user-db
NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE CONTAINER(S) IMAGE(S) SELECTOR
deploy/carts 1 1 1 1 6m carts weaveworksdemos/carts:0.4.8 name=carts
deploy/carts-db 1 1 1 1 6m carts-db mongo:3.2.11 name=carts-db
deploy/catalogue 1 1 1 1 6m catalogue weaveworksdemos/catalogue:0.3.5 name=catalogue
deploy/catalogue-db 1 1 1 1 6m catalogue-db weaveworksdemos/catalogue-db:0.3.0 name=catalogue-db
deploy/front-end 1 1 1 1 5m front-end 2886795275-4567-kitek01.environments.katacoda.com/root/front-end name=front-end
deploy/orders 1 1 1 1 5m orders weaveworksdemos/orders:0.4.7 name=orders
deploy/orders-db 1 1 1 1 5m orders-db mongo:3.2.11 name=orders-db
deploy/payment 1 1 1 1 5m payment weaveworksdemos/payment:0.4.3 name=payment
deploy/queue-master 1 1 1 0 5m queue-master weaveworksdemos/queue-master:0.3.1 name=queue-master
deploy/rabbitmq 1 1 1 0 5m rabbitmq rabbitmq:3 name=rabbitmq
deploy/shipping 1 1 1 0 5m shipping weaveworksdemos/shipping:0.4.8 name=shipping
deploy/user 1 1 1 0 5m user weaveworksdemos/user:0.4.4 name=user
deploy/user-db 1 1 1 0 5m user-db weaveworksdemos/user-db:0.4.0 name=user-db
NAME DESIRED CURRENT READY AGE CONTAINER(S) IMAGE(S) SELECTOR
rs/carts-153328538 1 1 1 6m carts weaveworksdemos/carts:0.4.8 name=carts,pod-template-hash=153328538
rs/carts-db-1968389027 1 1 1 6m carts-db mongo:3.2.11 name=carts-db,pod-template-hash=1968389027
rs/catalogue-114596073 1 1 1 6m catalogue weaveworksdemos/catalogue:0.3.5 name=catalogue,pod-template-hash=114596073
rs/catalogue-db-1956862931 1 1 1 6m catalogue-db weaveworksdemos/catalogue-db:0.3.0 name=catalogue-db,pod-template-hash=1956862931
rs/front-end-537982664 1 1 1 5m front-end 2886795275-4567-kitek01.environments.katacoda.com/root/front-end name=front-end,pod-template-hash=537982664
rs/orders-2365168879 1 1 1 5m orders weaveworksdemos/orders:0.4.7 name=orders,pod-template-hash=2365168879
rs/orders-db-3036494983 1 1 1 5m orders-db mongo:3.2.11 name=orders-db,pod-template-hash=3036494983
rs/payment-1968871107 1 1 1 5m payment weaveworksdemos/payment:0.4.3 name=payment,pod-template-hash=1968871107
rs/queue-master-2798459664 1 1 0 5m queue-master weaveworksdemos/queue-master:0.3.1 name=queue-master,pod-template-hash=2798459664
rs/rabbitmq-2262854187 1 1 0 5m rabbitmq rabbitmq:3 name=rabbitmq,pod-template-hash=2262854187
rs/shipping-2899287913 1 1 0 5m shipping weaveworksdemos/shipping:0.4.8 name=shipping,pod-template-hash=2899287913
rs/user-468431046 1 1 0 5m user weaveworksdemos/user:0.4.4 name=user,pod-template-hash=468431046
rs/user-db-1166754267 1 1 0 5m user-db weaveworksdemos/user-db:0.4.0 name=user-db,pod-template-hash=1166754267
@BenHall thanks. Is the config repository public?
It's part of Katacoda so it's only short lived. It is public at the time. Why do you ask? To reproduce?
To reproduce?
Yes. The set of services listed by list-images
is the one obtained from k8s, whereas the services used by release
are further filtered by what's defined in the config repo. Since the former is successfully finding sock-shop/front-end
and it is missing from the latter ('Found 0 services.'), the finger points at the config repo...
I gather from the subject of your issue that you think the problem is caused by k8s 1.6 - is there a way to check that this doesn't also happen with earlier versions?
The current Katacoda (https://www.weave.works/docs/tutorials/core/interactive/part-2-deploy-continuous-delivery/) is using v1.5 and works as expected.
The same repository with a Kubernetes v1.6 cluster fails
fluxctl release --service=sock-shop/front-end --update-all-images
Submitting release job...
Release job submitted, ID 97e450be-09e3-cfba-081a-1b84c57a0b08
Status: Complete.
Here's what happened:
SERVICE STATUS UPDATES
sock-shop/front-end skipped not found in repository
Took 10.98256s
A copy of the microservices-demo repo we use can be found at http://assets.joinscrapbook.com/weavecloud/weave-micro-0604.tar.
Would the fact that I'm using :latest be causing an issue? I noticed the image Kubernetes returns is 2886795275-4567-kitek01.environments.katacoda.com/root/front-end
Flux 0. is no longer supported. Please file a new issue if you are having difficulties with Flux 1., Ben.
On Kubernetes 1.6, everything looks to deploy correctly, but I'm unable to view the images. Is this a known issue or something with my configuration?