Joe,
  Hi, Mike here,  I am curious as to the hardware you use - I would like to 
also test on some arm chips - the only arm chips I have access to in the one in 
my iphone and the cortex based on in the PI3 
https://wiki.onap.org/display/DW/ONAP+on+ARM+Cortex-A53  - docker installs 
there but I stopped short of finishing the K8S install on a cluster of 4.  I 
understand that the docker install in the scripts is intel specific.  I assume 
we are running master right.  It would be nice if we could setup a page 
specifically for ARM64.

  For the scripts - make sure you are getting the latest from the JIRA under 
review (they are not merged to gerrit yet)
https://jira.onap.org/browse/OOM-710

  The prepull script (only for master) just parses all the values.yamls to get 
the docker tags and pull them before starting up all the containers (90 pulls 
take 20-240 min).  Are you able to login and pull any one of them manually - 
all the script does is pull 90 docker images in parallel - so if one works they 
all will (you will see random errors - all of us get these periodically against 
nexus3).

  Try just bringing up a single component like robot, then a larger one that is 
a leaf in the dependency tree like AAI.  This would validate your setup.
  ./createAll.bash -n onap -a robot
  Or 
  ./createAll.bash -n onap -a aai
  Of course after running ./createConfig.sh -n onap - which would pull down the 
config container from the oomk8s repo on dockerhub - it must finish with a "0/1 
completed" before you can proceed - see the script.

   There is a fully automated oom_rancher_install.sh that is pulled down by 
oom_entrypoint.sh (onap-parameters.yaml is only the simpler master version 
right now) - that you can use as a reference (retrofit for arm).

   To simplify things (take the distributed share across your 4 vm's out of the 
equation) - try installing robot or aai like above on a single collocated VM 
(rancher and oom on a single 16g vm) - just to very your config.

   However I see your big blocking issue is the AMD64 vs ARM64 based docker 
images - so you may need to reproduce the docker daily merge jobs of the CI 
system to build your own - I think we may need to seriously involve the Linux 
Foundation here.
 
  I am travelling right now - but I will add more info within 24 hours.
  /michael

-----Original Message-----
From: Joe Kidder [mailto:[email protected]] 
Sent: Sunday, March 4, 2018 20:05
To: [email protected]; Michael O'Brien <[email protected]>
Subject: Running ONAP on arm64 - First need to build ONAP containers for arm64 

Hello ONAP Enthusiasts,
  I am currently working on the Auto project in OPNFV, and one of our goals is 
to run ONAP on arm64 servers.
  At this point, I’m following the installation process described by Michael 
O’Brien in [1].
  I currently have a 4-server k8s cluster (1 master and 3 nodes) deployed on 4 
VMs an arm64-based OPNFV open stack deployment.
  I also have helm running on that cluster.

  When I start executing the steps in the “cd.sh” script described in [1], I 
can see that the pulled docker images are built for arch=amd64, so they don’t 
run.

  Can someone provide some advice/starting-hint to build the various ONAP 
components?
  In the meantime, I’ll start down the path of described here [2] at the “Build 
ONAP” partway down the page.

Thanks very much!
Joe Kidder
 
[1] 
https://wiki.onap.org/display/DW/ONAP+on+Kubernetes#ONAPonKubernetes-ExampleEndtoEndKubernetesbasedONAPinstallanddeployment
[2] https://wiki.onap.org/display/DW/Setting+Up+Your+Development+Environment
This message and the information contained herein is proprietary and 
confidential and subject to the Amdocs policy statement,

you may review at https://www.amdocs.com/about/email-disclaimer 
<https://www.amdocs.com/about/email-disclaimer>
_______________________________________________
onap-discuss mailing list
[email protected]
https://lists.onap.org/mailman/listinfo/onap-discuss

Reply via email to