Handling Errors from Blockchain

[saraswatpuneet]

Closes: #18

Details

• Check for the existence of txHash: this should always be there even if messages pallet failed (reason: capacity will be withdrawn as expected and low capacity related error is handled in publisher itself: read signed extensions)
• Check for success events: If capacity withdrawn found and message stored events are found then it is a complete success
• Check for error from module (messages pallet) : here based on error take action like, fail the job, re-queue or anything else
• Refactor parts to blockchain lib

[aramikm]

Added some comments. Please let me know if you might like to go over a couple of those errors to figure out what are the options to handle them.

[saraswatpuneet]

Here is how we are currently handling errors

1. if a tx is not found i.e. did not make it to block by the time we check, an error is thrown so that retry can trigger
2. if a tx is found and success events are discovered then we are good
3. if a tx is found and error is found for messages pallet then we take action to either pause the publishing or retry it, while i most cases the job is marked failed as we want to have a receipt of tx given it did consume capacity when failed too
4. Ideally in retry logic we queue the publish job and mark the tx receipt job as failed and once publishQueue gets the job done it will be re-queued in tx-receipt

[aramikm]

Looks awesome! I just added a few suggestions.