sync from 0 broken on mainnet/devnet (sometimes)

[alessiodf]

Devnet 2.6.0-next.10:

1|ark-core  | [2020-02-07 00:55:00.476] DEBUG: Delegate darkgalp (037997a6553ea8073eb199e9f5ff23b8f0892e79433ef35e13966e0a12849d02e3) allowed to forge block 611,446
1|ark-core  | TypeError: Cannot read property 'keysgroup' of undefined
1|ark-core  |     at MultiSignatureRegistrationTransaction.serialize (/home/ark/.config/yarn/global/node_modules/@arkecosystem/crypto/dist/transactions/types/multi-signature.js:54:60)
1|ark-core  |     at Function.serialize (/home/ark/.config/yarn/global/node_modules/@arkecosystem/crypto/dist/transactions/serializer.js:39:40)
1|ark-core  |     at Function.toBytes (/home/ark/.config/yarn/global/node_modules/@arkecosystem/crypto/dist/transactions/utils.js:11:40)
1|ark-core  |     at Function.serializeWithTransactions (/home/ark/.config/yarn/global/node_modules/@arkecosystem/crypto/dist/blocks/serializer.js:29:53)
1|ark-core  |     at Function.serializeWithTransactions (/home/ark/.config/yarn/global/node_modules/@arkecosystem/crypto/dist/blocks/block.js:67:40)
1|ark-core  |     at Function.fromData (/home/ark/.config/yarn/global/node_modules/@arkecosystem/crypto/dist/blocks/factory.js:37:42)
1|ark-core  |     at Blockchain.processBlocks (/home/ark/.config/yarn/global/node_modules/@arkecosystem/core-blockchain/dist/blockchain.js:316:64)

At which point sync stops until the process is manually restarted. Had to restart the node 5 times before it would continue. An identical problem also exists at height 368177.

If it is because of a bad peer sending a block with an invalid asset, at least Core should handle it more gracefully and not require a (repeated) manual restart.

Thanks for opening this issue! A maintainer will review this in the next few days and explicitly select labels so you know what's going on.

If no reviewer appears after a week, a reminder will be sent out.

[alessiodf]

Probably related to the difference in responses across peers, e.g.:
http://142.44.253.192:4003/api/transactions/9a3a919b2b9dde7ff91c531f928f4141ff22afc1a9185b98c1a8075f57025a29

vs

http://51.89.136.132:4003/api/transactions/9a3a919b2b9dde7ff91c531f928f4141ff22afc1a9185b98c1a8075f57025a29

(lack of signatures attribute)

[air1one]

Sorry for late answer, we are looking into this 😄 will keep you updated here. Thanks for reporting !

[ksemaev]

Yep, same for me, got it on 2 of my 6 nodes when resync from 0

[alessiodf]

Maybe in the event of an unhandled error during sync (which this is), it should try redownloading the block from a different peer. That way it will eventually hit a good peer and carry on, rather than bailing out and requiring a manual process restart.

[ksemaev]

I did a small loop for restarting in case of error, and it already made more them 100 cycles, and can't make it over this block. Any advices on how to discover and connect to good peers?

[faustbrian]

This should be patched once we release the next version on mainnet.

This issue has been closed. If you wish to re-open it please provide additional information.