Set back lower memory limits to have stack working by default on

[apupier]

che.openshift.io eclipse-che/che#13832

Signed-off-by: Aurélien Pupier apupier@redhat.com

What does this PR do?

What issues does this PR fix or reference?

[apupier]

tried on https://che.prod-preview.openshift.io , the workspace is starting. Unfortunately, I have a "dial tcp 172.30.45.163:3100: connect: no route to host" which doesn't seem to be related to this PR

[apupier]

Unfortunately, I have a "dial tcp 172.30.45.163:3100: connect: no route to host" which doesn't seem to be related to this PR

maybe related... can reproduce only with this one.

there are error 500 reported in console log

[apupier]

seems an erro unrelated to memory limits.

there is "/entrypoint.sh: line 10: /home/user/.bashrc: Permission denied" in log of the "maven" pod

[apupier]

this error is known and not related to my PR. @amisevsk will provide a fix soon

[amisevsk]

The entrypoint.sh issue will be resolved by #42

[amisevsk]

To confirm, starting a workspace from the updated devfile in this PR gives the error

Terminated at Jul 23, 2019 1:51:45 PM with exit code 137 (OOMKilled)

on the vscode-camel sidecar container.

[apupier]

updated to provide lower limits that allows each VS code extensions to start while keeping total lower than 3G

[amisevsk]

Briefly tested this devfile with the rc-4.0-SNAPSHOT build and found no specific issues. I worry that the lower limits might be hit in some case I didn't cover, but it seems to work for now.

Thanks @apupier !

[ibuziuk]

tested against prod-preview and workspace is started just fine without hitting quotas

[ibuziuk]

@l0rd @rhopp are you ok to merge ?

[ScrewTSW]

@apupier Hello. I've done some manual investigation myself as well. Here are my suggested values.

Maybe you can consider my proposal and possibly update this PR.

https://gist.github.com/ScrewTSW/892a1b8d3ca2a98f798f7adba0475959

[l0rd]

The choices are:

• merge this PR
• remove the camel springboot devfile before GA (because it doesn't work)

+1 to merge this considered that it's a 2 lines fix

[slemeur]

+1 to merge

…

On Wed, Jul 24, 2019 at 11:10 PM Mario Loriedo ***@***.***> wrote: The choices are: - merge this PR - remove the camel springboot devfile before GA (because it doesn't work) +1 to merge this considered that it's a 2 lines fix — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#46>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAMPTINVUWJV2VMJO6OIQ4TQBDAMTANCNFSM4IGESW5Q> .

-- Stévan LeMeur // Product Manager // Developer Tools // +336-87-11-27-55

[apupier]

@apupier Hello. I've done some manual investigation myself as well. Here are my suggested values.

Maybe you can consider my proposal and possibly update this PR.

can you explain the modification of the values? What are they improving more precisely?

I tried with these values and I cannot launch the "Tasks". There is an error in browser console:
task ERROR Can't get task launch configuration for label: run the services but cannot find anything in pod log explaining that.

[ScrewTSW]

@apupier Hello. I've done some manual investigation myself as well. Here are my suggested values.
Maybe you can consider my proposal and possibly update this PR.

can you explain the modification of the values? What are they improving more precisely?

I tried with these values and I cannot launch the "Tasks". There is an error in browser console:
task ERROR Can't get task launch configuration for label: run the services but cannot find anything in pod log explaining that.

That's strange. I've been running it on prod-preview starter-2a-cluster with no issues, running all tasks. I've modified the values based on the commands execution and demo app deployment and testing. I've looked into the cluster monitoring to see how much memory the runners consume and modified the values based on that.