José Alberto
José Alberto
I think both, time-in and time-out features can be used and combined for great flexibility :)
yep, already tried those workarounds, didn't work. notice I am on kubernetes, so I cannot "reboot node" I recreated it, anyway didn't work.
yes, x-pack is installed, but every premium feature is disabled (in ES and kibana). ES is a cluster, but kibana only have one instance note: using official ES docker images
I tried several methods: 1. using postStart hook -> got reported error 2. using a shell though kubectl exec -> same erro 3. building a custom image -> works as...
yes, but even restart doesn't works, at least in k8s context as the most similar thing is recreate pod
installing by hand using `kubectl exec` end then recreating the pod (using persistent volumes of course) doesn't works either
looks like, for that reason I suggest to explicitly mention it in the docs, so next one dont spent too much time "fixing" this :)
Do this works with OAS3 ?
in addition to @si-rob `lr: null` it's also needed but still, there is an issue with the models: `nboost OSError: Model name 'pt-bert-base-uncased-msmarco' was not found in model name list...
So the configured probes in helm are causing that `MissingQuery` exception, disabling the probes is workaround thou. Now Next error: ``` │ nboost E:Proxy:[pro:han:132]: │ │ nboost Traceback (most recent...