aleph-im/pyaleph

Large file uploads time out on v0.5.0-rc1

Closed this issue · 1 comments

odesenfans commented

Files in the GB range have been measured to take several hours to be downloaded by the IPFS daemon on a CCN. The retry mechanism rejects these messages on Postgres nodes. However, they end up being fetched on MongoDB nodes because the pin command is still issued to the IPFS daemon. As MongoDB nodes retry messages forever, they end up being processed. For now the on-chain sync mechanism makes the Postgres nodes retry the message but this won't be the case forever.

Possible causes:

  1. The IPFS protocol is too slow for large files (unlikely?)
  2. The IPFS daemon is overloaded with requests and is doing too many things at once.
odesenfans commented

Fixed in #392.