Skip to content

Retrieval failure from indexer: context canceled when successful but _fast_ #402

@rvagg

Description

@rvagg

Testing out some http failures from web3.storage that now seem to be passing, I'm guessing related to storacha/freeway#87 where the content hasn't quite been pushed out to the caching layer. But this is interesting behaviour from lassie:

go run ./cmd/lassie/ fetch -p --protocols http bafybeihmqcb3kjyib3jed5cudpvfyf7ydquh4rjykma34qxnmj6von66d4/6762.png
... [gobs of indexer candidates, repeating only 2 peer IDs many many times, one for bitswap and one for http (w3s)]
Received 9 blocks / 1.4 MiB...
Fetched [bafybeihmqcb3kjyib3jed5cudpvfyf7ydquh4rjykma34qxnmj6von66d4] from [QmUA9D3H7HeCYsirB3KmPSvZh3dNXMZas6Lwgr4fv1HTTp]:
        Duration: 560.943093ms
          Blocks: 9
           Bytes: 1.4 MiB
Retrieval failure from indexer: context canceled

This happened after running it a couple of times, note how fast the download is, so this is super speedy now it's cached in CF from web3.storage, and I think lassie trips over itself and we get a context cancellation in the indexer streaming results that we count as an actual failure. But we got the data we wanted, and it validated, so should be an overall success.

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions