how does fast registration work When I try giving it an s3 o

Question

how does fast registration work When I try giving it an s3 output like the docs `pyflyte register p project name d domain name output <s3 my s3 bucket raw data>` lt subbing in my own bucket I get `FileNotFoundError Errno 2 No such file or directory lt GIT REPO PATH gt s3 my s3 bucket fast serialize fast3e198a8e9dd654e746828c0ae929fce3 tar gz ` and when I don t feed in a ` output` it creates a new folder inside `my s3 bucket` When I run it from the UI I get `tar development fast3e198a8e9dd654e746828c0ae929fce3 tar gz Cannot open Not a directory`

Yee · Answer

Because users update code a lot more frequently than they update the image Flyte has this notion of fast register Think of it as a patch to your image before running in practice flytekit will up your code from what it detects is the source upload thar archive to the blob store configured with Admin by a signed url for upload via the service and then a command to the task s normal command That command at run time will that archive back from the flyte configured blob store and it to the root of your container s image thus saving the user a build cycle But generally you should think of the act of registration as a two step process The two steps are compilation which is the act of parsing the Python code and creating IDL protobuf based objects that represent them and registration the act of shipping those proto objects to a Flyte backend In fact there are still separate commands for the two steps Instead of running `pyflyte register` you can still run `pyflyte package` and then `flytectl register` Here s a that describes the difference between them

Yee · Answer

cc < Ankit Goyal> if this helps

Yee · Answer

the ` output` switch you mention is meant for local use that s where the proto artifacts are saved in between those two steps

Yee · Answer

regardless of whether or not you specify an output it should create a new s3 file holding the contents of the archive

Laura Lin · Answer

why do I get this error tho `tar development fast3e198a8e9dd654e746828c0ae929fce3 tar gz Cannot open Not a directory` when I try running a task after doing pyflyte register ```subprocess CalledProcessError Command tar xvf development fast3e198a8e9dd654e746828c0ae929fce3 tar gz C development returned non zero exit status 2 ```

Laura Lin · Answer

wait taking another look ` d` is mapped to domain and destination dir

Laura Lin · Answer

ok I got past the errors and it seems to run now but the code that was run isn t changing but the version did get bumped And I can see that the container args do call `pyflyte fast execute` ``` task original docker task def fast serialize test task test int int gt str return str test int + 1 task fast serialized def fast serialize test task test int int gt str return str test int + 1 ``` When I look at the file inside the tar gz I see the new version of the task

Yee · Answer

can you download the tar file locally and see if it indeed has the updated code

Laura Lin · Answer

yea I did and it does

Yee · Answer

is it unpacking to the wrong place

Laura Lin · Answer

where is it supposed to unpack to

Laura Lin · Answer

in my docker I have the code copied over to ` lt repo gt src` so I set the destination dir to ` lt repo gt src` also

Yee · Answer

< Eduardo Apolinario eapolinario >

Eduardo Apolinario (eapolinario) · Answer

< Laura Lin> one thing to confirm here is what you re setting `WORKDIR` in your dockerfile

Laura Lin · Answer

yes its set in the docker

Eduardo Apolinario (eapolinario) · Answer

Cool thanks for confirming Did you have a chance to use the latest update in that branch to kick off an execution

Laura Lin · Answer

I ll do it ~10 min