1

I am trying to install the Boreas release of Acumos on a Ubuntu 18.04 virtual machine (6 CPUs, 16GB RAM and 200GB Disk).

I have setup Docker and Kubernetes (Minikube) and have run this to install Acumos Boreas:

All commands have been issued as root (sudo -s)

Get the install setup

1) git clone https://gerrit.acumos.org/r/system-integration

2) cd system-integration/AIO

3) bash setup_prereqs.sh k8s acumos.tele.no root

4) bash oneclick_deploy.sh

First time I tried this it exited with failure "oneclick_deploy.sh: line 90: helm: command not found"

5) snap install helm --classic

Trying again

6) bash oneclick_deploy.sh

Now the install script complains (the password is for reference purpose, i.e. it is not what is shown in the real log entry):

setup_mariadb:68 (Fri Jul 19 12:31:40 UTC 2019) Mariadb server is not yet accepting connections from 193.156.17.73 + sleep 10 + mysql -h 193.156.17.73 -P 30001 --user=root --password=2a9be1c1-5347-42d9-9f43-410eed97915b -e 'SHOW DATABASES;' ERROR 1045 (28000): Access denied for user 'root'@'172.17.0.1' (using password: YES) + i=31 + [[ 31 -gt 30 ]] + fail 'MariaDB failed to respond after 5 minutes' + set +x

fail:42 (Fri Jul 19 12:31:50 UTC 2019) MariaDB failed to respond after 5 minutes ++ fail ++ set +x

fail:42 (Fri Jul 19 12:31:50 UTC 2019) unknown failure at main 282

Any tips on what might be wrong? The only strange thing I see in the Kubernetes Dashboard is a warning in the acumos-mariadb pod event log:

pod has unbound immediate PersistentVolumeClaims

Arne Munch-Ellingsen
  • 37
  • 4
  • I don't think 16GB is enough RAM to run all the services. As for maria DB, did you check if the container has been launched? A missing persistent volume could cause it not to start; too little RAM could cause the start to be massively delayed. Please open a Jira issue at https://jira.acumos.org/ – chrisinmtown Jul 19 '19 at 14:31

1 Answers1

0

One update to the process you followed, per https://docs.acumos.org/en/boreas/submodules/system-integration/docs/oneclick-deploy/user-guide.html#kubernetes-based-deployment, is to call the script 'system-integration/AIO/acumos_k8s_prep.sh' rather than 'setup_prereqs.sh' directly. 'setup_prereqs.sh' is called by 'acumos_k8s_prep.sh' once it has prepared things outside the scope of 'setup_prereqs.sh', e.g. helm install. That is why you got the Helm error.

The MariaDB error is also likely related to not running 'acumos_k8s_prep.sh', since that script kicks off MariaDB deployment using a Helm chart.

Bryan Sullivan
  • 71
  • 3