This message was deleted.
# ask-for-helps
Slackbot
09/27/2022, 12:12 PMThis message was deleted.
👀 1
🚀 1
🍱 1
a
Aaron Pham
09/27/2022, 12:58 PMHi @Judah Rand, you can mount
bento_configuration.yamlBENTOML_CONFIGj
Judah Rand
09/27/2022, 1:25 PM@Aaron Pham That doesn’t help me when deploying with
bentoctlJudah Rand
09/27/2022, 1:26 PMOr do I have to pass
BENTOML_CONFIGbentoctlJudah Rand
09/27/2022, 1:26 PMIf so that doc does not make that clear at all given that
bentomlbentoctla
Aaron Pham
09/27/2022, 1:57 PMyou can pass in the environment variable. bentoctl is just a tool to deploy bentoml right. So I believe BentoML environment should just work as expected.
j
Judah Rand
09/27/2022, 2:51 PM@Aaron Pham This 100% does not work.
Judah Rand
09/27/2022, 2:52 PMIf I do
BENTOML_CONFIG=$PWD/bentoml_configuration.yaml bentoml containerize <PLACEHOLDER>:latestbentoml containerize <PLACEHOLDER>:latestJudah Rand
09/27/2022, 2:52 PMTherefore, the config can’t be any different.
Judah Rand
09/27/2022, 2:52 PMThere seems to be no way to pass static config into a Bento container image…
Judah Rand
09/27/2022, 4:04 PMThis seems like a pretty significant oversight 🤔 It means that AutoBatching is basically unusable as there is no way to override the
max_latency_ms=10000Judah Rand
09/27/2022, 4:05 PMIt seems to me the only place that the
BENTOML_CONFIGbentoml servea
Aaron Pham
09/27/2022, 4:18 PMyou have to mount the environment variable to the container
Copy code
docker run --env BENTOML_CONFIG=/home/bentoml/bentoml_config.yml -v /path/to/config:/home/bentoml/bentoml_config.yml ...j
Judah Rand
09/27/2022, 4:19 PM@Aaron Pham And how do I do that in Google Cloud Run?
Judah Rand
09/27/2022, 4:21 PMI do understand exactly what the issue is now (having spent a decent chunk of the afternoon experimenting) and realize that by editing the
<http://main.tf|main.tf>bentoctla
Aaron Pham
09/27/2022, 4:22 PMcc @jjmachan probably have a better idea about google cloud run
b
Bo
09/27/2022, 4:38 PM@Judah Rand you are right. This is a very big oversight. We will try adjust this in the next release for google cloud run and to other deploy target afterwards
Bo
09/27/2022, 4:38 PMAre you using it in production right now?
j
Judah Rand
09/27/2022, 4:55 PMAre you using it in production right now?Not atm but aiming to in the very near future assuming we can get it to work. The aim is to have this be the ‘template’ for how we’d like to deploy internally facing models going forwards and potentially customer facing models further out.
👍 1
s
Sean
09/28/2022, 2:54 AMThe
v1.0.6