Welcome to the Linux Foundation Forum!

Lab 15.1 - MariaDB package takes a very long time to load

sdalziel
sdalziel Posts: 17
edited May 2018 in LFS258 Class Forum

Hi all,

I worked through 15.1 twice but even after 6hours the pod was never successfully started.

Could I get some help on where to begin troubleshooting an issue like this, please?

My environment is 2 instences of Amazon Web services EC2-t2-Large. Is this enough processing power for this Pod along with the others we have running over the course?

Also, let me know what more information I can provide.

I've been using the script command to keep a log of my STDIN/STDERR messages.

 

Cheers,

Sean

Comments

  • chrispokorni
    chrispokorni Posts: 2,155

    Hi Sean, your instance sizes are sufficient, just as recommended in lab 3. 

    The first time I went thru lab 15 I also had trouble getting everything up and running. It took a second time and paid a lot of attention at every step to get the lab running as expected. I completed the lab on GCE, but I do not expect GCE and EC2 to behave differently. I also created a firewall rule to open all ports for all traffic, no VPC, no SG or NACL of any kind.

    Did you find anything unusual in the logs? Anything to indicate a hang, timeout, etc?

    -Chris

  • scullyt
    scullyt Posts: 1
    edited June 2018

    Hi Sean,

    Are your mariadb pods ever starting or remainig in ContainerCreating state?  I found that the Helm chart deploys 2 pods (master/slave) and that both try to claim a PV.

    You'll be able to see if that's the case using 'kubectl get events' in your namespace.

    Using 'helm inspect' I found that setting the 'persitence.enabled' for both master and slave was required, like this:

    helm --debug install stable/mariadb  \

    --set master.persistence.enabled=false --set slave.persistence.enabled=false

    Hope this helps.

    Tony

     

  • chrispokorni
    chrispokorni Posts: 2,155

    Hi Tony,

    Thanks for the tip. After reading your post I took another swing at this lab and realized that things have changed a bit since I completed it the first time. The persistence.enabled has to be set for both master and slave separately, just like you mentioned in your post. I believe this change is a result of the mariadb chart getting upgraded from version 2.1.3 to version 4.2.2 as displayed by the output of 


    helm search database

    command in step 13. In the earlier version it was enough to set it once, but in the current version the documentation at 

    https://github.com/kubernetes/charts/tree/master/stable/mariadb

    shows 2 separate config parameters. This change should be reflected in the next course update release. 

    Regards, 

    -Chris

Categories

Upcoming Training