Does a step fail silently if the
Does a step fail silently if the returned amount is more than the 1MiB? Its the only conclusion i can reach as to why my step is constantly restarting.
10 Replies
Hi, can you send me an instance id and workflow name so that I can investigate?
And please run the /link command @SODaLC
Should be linked now
Workflow name is: league-patch-indexer
Instance ID is bfa4f874-bfdd-401f-824d-ed9cc4b0b5f7.
I batch started a few which i had to terminate each as i discovered this issue.
I am going to investigate and will get back to you, if you have the step name that also expedites things a lot
The step was "Step Three: Fetch all data for version from DDragon"
This step had 5 internal Steps all worked and returned data. But when it came to returning to the main step, the step kept restarting.
Sorry, I am now realising the one I sent was wrong. When i mentioned this i forgot to say that I found the silent fail and step restarting locally before deploying to CF. Ill deploy and run a instance to show this issue then send that one across.
Okay, right. This is the ID: 04d2feca-c8c0-42d7-b290-a9529f16f973 for Workflow: league-patch-indexer
On CF, it does indeed state that the output was too large. So this is more a Wrangler issue, that does not say this during development. Apologies, for wasting your time on this.
No worries we will get that fixed on wrangler as well
here what it does on local in console
Thank you for your time.
Hey @SODaLC , what version of wrangler are you using?
You should have seen
✘ [ERROR] Uncaught Error: string or blob too big: SQLITE_TOOBIG in your logs, unless you are catching errors at the top level of your workflowGlobal: 4.41.0
Project: 4.20.0
I use global wrangler over project wrangler. Is this where the issue could lie?
When you say top level of the workflow, do you mean inside the run function?
If so, yes, there is a try catch block encompassing all the steps. I am outputting any errors to console in my catch block.
These versions are both fine, so no issue here
Yes, inside the run function.
If your catch block is only logging the errors, it should be logging this error
string or blob too big: SQLITE_TOOBIG. Unless you have additional logic in there.
Can you confirm that you deployed the exact same version of that code and the deployed one did in fact log the error?Yes, the current code on CF and Local are currently the same. But i think i'm starting to see the confusion now.
So, when I ran locally the first time yesterday and discovered it, I immediately terminated the command. I just ran it again now to make see it fail after the 5 retries, only then does the error appear when it hits my catch block. But, because I terminated after seeing it loop the first time, that's why I thought it was silently failing and made the comment yesterday.
To further analyse this:
The Workflow Dashboard (Workflow: lol-patch-indexer, Instance ID: 7cf40be2-4916-4f77-a8a7-c94590ea6b66, Step Three) shows the step error out and shows the error reason.
The Worker Observability (Request ID: 3RRT880X42OGXNPY) mimics what I am now seeing locally.