Polygon POS Problem

[sidohin-felix]

When launching the latest build of erigon built from main the following happens when syncing with the polygon pos flags enabled after a certain time.

pos sync store failed: nonsequential block in bridge processing: 66842880 != 66842848"

This leads to erigon to crash and restarting it works for a few hours then the error comes back again.

Any suggestions on this?

All I found is the following in the source:

expectedNextBlockNum := lastProcessedBlockInfo.BlockNum + s.borConfig.CalculateSprintLength(blockNum)
		if blockNum != expectedNextBlockNum {
			return fmt.Errorf("nonsequential block in bridge processing: %d != %d", blockNum, expectedNextBlockNum)
		}

I am not exactly sure how this works for Polygon, but my guess would be it might that we should have <= rather than not equals since we were faster.

[taratorio]

@sidohin-felix we released a new flow for polygon pos for preparation for beta which requires re-sync from scratch (most preferable). Alternatively you can try stopping erigon, then rm -rf datadir/chaindata && rm -rf datadir/heimdall && rm -rf datadir/polygon-bridge and starting again - if that doesn't help please try re-syncing.

[sidohin-felix]

@taratorio so I tried re-synching a few times before but the error always comes back, I also thought this was related but to no avail. I applied my strange suggestion but I used a strict less than and for now everything seems stable (even though I think the solution is wrong). I can always revert to the original code and try the the rm -rf solution...

[taratorio]

@sidohin-felix please upload your full logs file with --log.console.verbosity=debug without your change with latest main once you get the error again. We haven't seen this error in our tests. The check is correct as it is and if you are seeing an error it means that something else is going wrong (for which I need your entire log file with debug verbosity to troubleshoot).

[sidohin-felix]

Ok I will revert to the original code and set to the log level - and yes it crashed again but with a different error. I will also provide all the build details.