-
-
Notifications
You must be signed in to change notification settings - Fork 174
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Would like to get this working with OpenShift Origin 3.7 #27
Comments
I should add that I don't know ansible at all. I was unsure what the instruction at the end about retrying with Also, the end of the Installer Status section has "This phase can be restarted by running: playbooks/byo/openshift-cluster/service-catalog.yml". Perhaps I should run |
I tried adding openshift_repos_enable_testing=true to inventory.template.cfg file as suggested in response to openshift/openshift-ansible#6086. That did get past the RBAC error, but I then saw: FAILED - RETRYING: Verify that TSB is running (1 retries left). Several thing occur to me:
Additionally, the advanced installation docs for OpenShift Origin indicate that one is supposed to set openshift_template_service_broker_namespaces if enabling the template service broker. See https://docs.openshift.org/latest/install_config/install/advanced_install.html#configuring-template-service-broker. But I tried adding Now, I'm going to try disabling the service catalog and template service broker with: openshift_enable_service_catalog=false I'm also going to explicitly set the ports with: |
That did not work either, but I now think I had added the new variables in the wrong part of the file, under [nodes] instead of under [OSEv3:vars]. I will retry. |
Unfortunately, even when I put the variables in the right place, the TSB running could not be verified. However, good news is that I was able to install OpenShift Origin 3.7 by disabling the service catalog and TSB with: openshift_enable_service_catalog=false One other note for you: I think you should technically include "etcd" under [OSEv3:children] at top of the inventory template. |
Hmm OK cool I'll take a look at this, thanks for sharing @rberlind! |
No problem. Thanks for putting together this repo. It was very helpful to me. |
Hi, did anyone solve this issue so far? The service catalog is a viable feature… |
Hi, Are you guys still having this problem? Release 3.7 worked for me. |
I have not used recently and had worked around it.
Roger
…On Tue, Feb 13, 2018 at 3:20 PM, Vang Nguyen ***@***.***> wrote:
Hi,
Are you guys still having this problem? Release 3.7 worked for me.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#27 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AWhd9PihVC-zZpAf14_3Njxy9nVfbSX0ks5tUe6MgaJpZM4Qvuki>
.
|
I'm on holidays at the moment and have no access to the git repository of my company. But I've worked around this issue by making a small change to the playbook. Please ask me again in the mid of march if you need some infos. |
Looking at this as well at the moment, note to self: https://docs.openshift.org/latest/install_config/configuring_aws.html#aws-cluster-labeling May need to update the labelling logic introduced in #33 Also check this for notes on dynamic aws tag names (particularly the limitations for how we can manage this in terraform): |
Hi @yves-vogl @rberlind @mtbvang, Can you let me know the changes you had to make to get this to work? At the moment, when I try to install 3.7, I always get this issue:
My current work-in-progress branch for this is here (I've opened a PR to make it easy to see the changes): The key changes so far are:
That's basically it. I've found the following issues which seem to be potential causes:
I've attempted the following workarounds:
So far no luck. Any pointers would be super helpful! |
Well, I cheated by disabling some things. Specifically, I was able to install OpenShift Origin 3.7 by disabling the service catalog and TSB with: openshift_enable_service_catalog=false I had also set openshift_repos_enable_testing=true in the inventory.template.cfg file. |
I also faced same issue around service catalog. I disabled below parameters as mentioned by @rberlind in host inventory file: openshift_enable_service_catalog=false And the deployment of OpenShift Origin 3.7 was successful: INSTALLER STATUS ******************************************************************************************************************************************************* My question: Is service catalog a mandatory feature for OpenShift environment? |
@rberlind I've tried this just now but no joy! Any chance you can share your inventory so I can take a look? @sumitshatwara You should be fine - the service catalog is an optional feature and you can test volumes without it, let me know how it goes!! |
I just applied my playbook again for openshift 3.7 for the modified version that runs on centos instead of rhel and it worked. I did get it running on rhel before making the changes to centos. I'm not hitting the issues that everyone else is so I'm a bit confused. The code I'm working with is in the centos branch in my fork. Lower down I list my development setup and the few changes that I made in commit mtbvang@8445f08. Git did a few weird things with the image files and I ended up committing them again, maybe because I'm working in a vagrant VM. Here's the output from the run and address of the cluster https://54.93.200.118.xip.io:8443 The only other thing I can see is I make some ssh key changes to name the key to terraform-aws-openshift. I work in a vagrant VM and this key is copied into the guest VM where I run terraform from. Below is more details about my vagrant setup. These are the differences I see between what I've done:
oc version on master
In my fork my current setup is in the centos branch. I've added a vagrant folder which has a vagrant file that spins up a centos development VM that installs the aws client tool version 1.14.36, terrafrom 0.11.3, and openshift client v3.7.0. You'll need to build the vagrant box using packer. There's a packer task in the vagrant/build.gradle file that can be run with with gradle wrapper from your host:
Once packer finishesf
I hope this helps in figuring this out, or ping me if you need any more information. |
I'm also seeing the issue with 3.7 version. |
Hi @dwmkerr, I did this back in December, but I believe the key changes I made were the following: inventory.template.cfg:
install-from-bastion.sh:
|
@rberlind hi, do you have tried the latest version now ? for me now , same version ,same problem |
I have not @junsionzhang . I have started working with this again, but have been creating a quite different version in which I trigger the ansible-playbook and all other installation steps with terraform remote-exec provisioners. I have not put any of this in Github yet. |
Suggesting to bypass 3.7 for those who are not tied to a particular version and just want to build a working cluster on a supported version of OpenShift. 3.7 seems to have a number of packaging issues, and the 3.9 release created additional problems for the 3.7 install. Things to change in the 3.7 branch here so that it can be used for the 3.9 install (amazingly, just a few):
|
Interesting that you mention the need to add "openshift_release=v3.9", @stanvarlamov. I just hit this the day before yesterday when I started getting errors about short_version 3.9 not being valid when using the 3.7 version of openshift-ansible. To keep using that version, I had to set openshift_release=v3.7. I also noticed that the documentation suggested using "openshift_deployment_type" instead of "deployment_type" and also made that change. By the way, another change that should be made is that the provisioning of the aws_instances should use "vpc_security_group_ids" instead of "security_groups" so that subsequent applies will not trigger destroy/create against the EC2 instances. See https://www.terraform.io/docs/providers/aws/r/instance.html#security_groups. |
I think in the spirit of this repo being an excellent source of procedures to get a working OpenShift version up on AWS with a basic configuration - we should move the master branch to 3.9 as that has been released recently. I find the 3.9 install process much improved compare to 3.6 and 3.7, and overall 3.9 features and look&feel more appealing. That is, basically, my suggestion. |
@stanvarlamov I think you're right on this one. A few months back I forked this project to make a multi-master setup using Centos instead of RHEL. Got all that working and noticed you guys added dynamic pv support so i merged with mine, i should mention as well before this i was using 3.7 just fine with few issues. After the merge a lot went wrong for me which i discovered you guys had already been through, some seemingly random stuff was going wrong too, but ended up here in this thread. I switched to 3.9 release instead and I got a lot further. I managed to get this project working with Centos, 2 Master nodes, 3 compute nodes (though not in ideal setup due to budget constraints), OpenShift 3.9 with Metrics and dynamic PV support. Still testing to make sure everything works ok. But can do a PR if you want to take a look, though i may have changed too much as its very specific to my needs. |
@Spengreb OpenShift 3.9 appears to be much faster than 3.6-7 and more stanble; PV resize also seems to be an important feature available. Centos, metrics and logging work as a 1-click install - pretty amazing, considering the amount of time it took to work through 3.6 Ansible bugs and inconsistencies. Dynamic PV based on EBS is a sore point, through. It kind of assumes you are single-AZ, which defeats the purpose of HA in a multi-AZ setup, so I don't consider that a production-ready feature at this point. But, overall, I think that 3.9 is really a game changer. Highly recommended. |
@stanvarlamov @Spengreb yep agreed! Will start on the 3.9 setup shortly. 3.7 has been a total pain to get working, so I'm all in favour of skipping it for now, if for some reason someone really needs 3.7 we can always go back and try again when some more of the ansible issues are sorted, but for now 3.9 sounds like a sensible option! |
you can close. |
When I try to use release-3.7 or release-3.7.0day by specifying them in the git clone command inside install-from-bastion.sh, I end up getting error at the end of the Ansible run:
TASK [template_service_broker : Reconcile with RBAC file] **********************
fatal: [master.openshift.local]: FAILED! => {"changed": true, "cmd": "oc process -f "/tmp/tsb-ansible-keZijh/rbac-template.yaml" | oc auth reconcile -f -", "delta": "0:00:00.285904", "end": "2017-11-29 12:45:42.125009", "failed": true, "rc": 1, "start": "2017-11-29 12:45:41.839105", "stderr": "Error: unknown shorthand flag: 'f' in -f\n\n\nUsage:\n oc auth [options]\n\nAvailable Commands:\n can-i Check whether an action is allowed\n\nUse "oc --help" for more information about a given command.\nUse "oc options" for a list of global command-line options (applies to all commands).", "stderr_lines": ["Error: unknown shorthand flag: 'f' in -f", "", "", "Usage:", " oc auth [options]", "", "Available Commands:", " can-i Check whether an action is allowed", "", "Use "oc --help" for more information about a given command.", "Use "oc options" for a list of global command-line options (applies to all commands)."], "stdout": "", "stdout_lines": []}
to retry, use: --limit @/home/ec2-user/openshift-ansible/playbooks/byo/config.retry
This seems related to openshift/openshift-ansible#6086.
I tried fixing the commit to 56b529e (which someone on that ticket said fixed the problem) by running
git checkout 56b529e
after thegit clone
command, but I got the same error.Can anyone suggest a workaround to get this working with OpenShift Origin 3.7? The problem is not with Terraform itself, but with the openshift-ansible code.
The text was updated successfully, but these errors were encountered: