error when adding custom runtime in catalog

[#1]

Hi,  

I am trying to add a custom runtime from gitlab private docker registry , and I am getting this error

"The provided full version does not start with the short version concatenated with maintenance version."

it might related to the path... 

I am trying to follow https://docs.cloudera.com/machine-learning/cloud/runtimes/topics/ml-distribute-the-runtime-image.htm...

are there rules for the 'user' ?

gitlab registry has some '/' , but I make sure that the last fields (sep by / ) are user , image-name:tag

any idea?

thanks

[#2]

I tried with  

registry/cdsw/image-name:tag

following this AMP example

https://github.com/cloudera/CML_AMP_Structural_Time_Series

but it does not work

[#3]

is the *host* in https://docs.cloudera.com/machine-learning/cloud/runtimes/topics/ml-add-docker-registry-credentials-...

the same as <company-registry>? thanks

[#4]

tried with my CDP/CML user-id and does not work

[#5]

tried the gitlab user (same of secret) and it does not work

[#6]

setting tag to 

com.cloudera.ml.runtime.full-version 

does not work

[#7]

setting tag to

short version concatenated with maintenance version, by a '-' does not work

[#8]

setting tag to

short version concatenated with maintenance version, by a '.' does not work

[#9 · Accepted solution]

sorted it out! 

the full version is actually taken from  the tag in the base image pulled in the docker file

e.g. in:

FROM docker.repository.cloudera.com/cdsw/ml-runtime-jupyterlab-python3.6-standard:2021.02.1-b2

it's

2021.02.1

so,  the following lines must look like 

ENV ML_RUNTIME_EDITION="Telnet Edition" \
ML_RUNTIME_SHORT_VERSION="2021.02" \
ML_RUNTIME_MAINTENANCE_VERSION="1" \

[#10]

btw, 

also it does not care about the user 

(for instance, no need to provide user as gitlab user, etc...)