Closed xinaxu closed 3 years ago
Thank you for the bug report.
I think the underlying cause is that on restart we are double-counting sent / received blocks in the data transfer layer. I've opened an issue here: https://github.com/ipfs/go-graphsync/issues/141
Closing since being tracked in graphsync#141
Describe the bug First, there have a been a few bug reports related to data transfer stuck so I'm not repeating that here. The issue my miner experienced is when I restart the data transfer using
lotus-miner data-transfers restart XXX
. The number of bytes transfered will increase however there is no network activity.To Reproduce Steps to reproduce the behavior: Assume there is a data transfer channel for storage deals that has transfered 321MiB and stuck.
lotus-miner data-transfers restart XXX
whereXXX
is the channel ID listed inlotus-miner data-transfers list
watch lotus-miner data-transfers list
, the bytes transfered for restarted channel will keep increasing till 642MiB and stuck againnload
to monitor network usage shows no obvious traffic increase while the bytes transfered keep increasingExpected behavior Either the data transfer should begin with real network traffic, or the data transfered should not increase.
Screenshots If applicable, add screenshots to help explain your problem.
Version (run
lotus version
): lotus version 1.3.0+git.19d457ae5.dirtyAdditional context https://filecoinproject.slack.com/archives/C01AZP8BKRQ/p1608309045337100?thread_ts=1608251272.292100&cid=C01AZP8BKRQ @dineshshenoy