iri
iri copied to clipboard
iotaledger
The node fails to fetch milestone transactions out of the db and stalls out
Bug description
The milestones that have been coming in are getting solidified and claim to have been stored correctly into the db, but when the latest milestone tracker is running, it fails to pull that transaction back out of the db, as a result the node stalls out. The issue is likely some sort of a race condition in the milestone solidifier. When it does the isSolid check, it returns true, then stops iterating through them.
It appears that the milestone isn't stored correctly so that it isn't recognized as a milestone again when being fetched again.
This picks up when restarting the node, so maybe the logic differs when the node is being restarted and the whole bundle is fetched correctly.
Concluded as investigation of #1447 (find the right issue)
Should be done in parallel with #1447
@DyrellC should help with investigating this.
🙂 🍶
IRI version
1.7.0
Steps To Reproduce
Ain't easy
- Start up a node
- Pair it with a node that is significantly ahead in milestones
Expected behaviour
What should happen.
Actual behaviour
What really happened.
Errors
Paste any errors that you see.
We decided we first need to investigate when this triggers nowadays, and if it at all does via scenarios for which this mechanism was conceived. In #1435
when #1447 will be done I guess this can be closed Putting it in the icebox so it won't interrupt us
