jonkristian•2mo ago

My flows stopped working after updating windmill

I updated windmill today, running on coolify and it appears that many of my flows stopped working. Looks like something to do with package cache or lock file, any pointers on how to proceed? I am seeing stuff like :

TypeError: Object is not a constructor (evaluating 'new import_woocommerce_rest_api.default({
    url: resource.url,
    consumerKey: resource.consumerKey,
    consumerSecret: resource.consumerSecret,
    version: resource.version,
    queryStringAuth: resource.queryStringAuth
  })')

TypeError: Object is not a constructor (evaluating 'new import_woocommerce_rest_api.default({
    url: resource.url,
    consumerKey: resource.consumerKey,
    consumerSecret: resource.consumerSecret,
    version: resource.version,
    queryStringAuth: resource.queryStringAuth
  })')

And

ExecutionErr: exit code for "deno run": 10, last log lines:
@supabase/supabase-js@2.14.0
error: Integrity check failed for remote specifier. The source code is invalid, as it does not match the expected hash in the lock file.
  Specifier: https://esm.sh/@supabase/supabase-js@2.14.0
  Actual: c277fa7166609cff04bb767940520ccbc912e801e24094e7175fcf83c8060d20
  Expected: dd0e88084937586b0a66acbe8aae117b019c49193a4403d95a32da1712c34def
This could be caused by:
  * the lock file may be corrupt
  * the source itself may be corrupt
Investigate the lockfile; delete it to regenerate the lockfile or --reload to reload the source code from the server.

ExecutionErr: exit code for "deno run": 10, last log lines:
@supabase/supabase-js@2.14.0
error: Integrity check failed for remote specifier. The source code is invalid, as it does not match the expected hash in the lock file.
  Specifier: https://esm.sh/@supabase/supabase-js@2.14.0
  Actual: c277fa7166609cff04bb767940520ccbc912e801e24094e7175fcf83c8060d20
  Expected: dd0e88084937586b0a66acbe8aae117b019c49193a4403d95a32da1712c34def
This could be caused by:
  * the lock file may be corrupt
  * the source itself may be corrupt
Investigate the lockfile; delete it to regenerate the lockfile or --reload to reload the source code from the server.

25 Replies

rubenf•2mo ago

esm.sh may have changed the source for those urls and we updated deno to deno 2, so those 2 probably ended up having the lockfile not matching anymore your best path is to re-generate the lockfiles, you can do this by adding a space character and redeploying your flows/scripts.

jonkristianOP•2mo ago

Will removing /tmp/windmill/cache also work? inside the worker container Thanks for the quick reply by the way!

rubenf•2mo ago

no the lockfiles are in the db

jonkristianOP•2mo ago

I tried but it is still failing with the same. I added some comments and spaces to the scripts inside the flow and re-deployed to trigger a regenerate, but didn't seem to help. It seems that it worked directly on a script, but the flow itself is still failing. I also added a new temporary step to the flow to make sure it has changed, but still the same error.

rubenf•2mo ago

you need to modify every script and inline script that throw that error the lockfile granularity is at the step level, not the flow level

jonkristianOP•2mo ago

Hm, I've modified all scripts in this particular flow that i am seeing this on.

rubenf•2mo ago

and you redeployed the flow?

jonkristianOP•2mo ago

yes, i hit deploy on the flow after modifying

rubenf•2mo ago

I'm not sure how but you must have missed a step

jonkristianOP•2mo ago

Let me double check everything again

jonkristianOP•2mo ago

I've changed the query script, the get skus, the supabase upsert and also the discord inputs, am i missing something?

jonkristianOP•2mo ago

I'm seeing this Lock summary, which i assume is something new in windmill, here's a more full screenshot of that:

rubenf•2mo ago

lockfileVersion 1, that lockfile look old did you re-deploy the workspace script?

jonkristianOP•2mo ago

i am doing all of this inside the flow, but should i edit scripts directly and re-deploy them first, and then edit the flow and re-deploy?

rubenf•2mo ago

you only need to redeploy the script unless you pinned the hash of that script in your flow in which case you need to unpin it

jonkristianOP•2mo ago

running test inside that specific script seems to show that it works, but if i hit deploy and run it fails.

rubenf•2mo ago

it fails with what error?

jonkristianOP•2mo ago

I also see that info at the bottom, have not tried the nobundling.

jonkristianOP•2mo ago

adding //nobundling seems to work though

rubenf•2mo ago

then your flow should work now

jonkristianOP•2mo ago

Just tested and yes, that made the flow work again. But this probably doesn't fix the core issue I'm having?

rubenf•2mo ago

what is your core issue?

jonkristianOP•2mo ago

If i remove //nobundling i am seeing the original error. But is it ok / should i be using //nobundling I guess is my question.

rubenf•2mo ago

yes, bun is unable to bundle your script so it will be slower, but not atrociously slow and the issue is with bun, you can reproduce with bun build locally it's a new optimisation, before we didn't bundle on deploy so yes you can add //nobundling without worrying

jonkristianOP•2mo ago

Ah thanks! I will try and do some test locally too. This flow just runs once per 30 mins to check for new product changes in woocommerce so should not be any problem then. Thank you for your time and help! I really appreciate it.

My flows stopped working after updating windmill

Did you find this page helpful?