getSyncInfo returns suspicios numbers for chainBestBlock - API v0.1.18

[theotniel]

API.getNet().syncInfo() returns unexpected values - i.e. the reported chainBestBlock is too low.

These are the numbers reported by the Java API:
Syncing: true, chainBest: 405384, netBest: 405785

These are the numbers reported by the web3 API:
aion-127.0.0.1:8545> eth.syncing
{ currentBlock: 405785,
highestBlock: 405384,
startingBlock: 400437 }

These are the last lines in the kernel's logs reffering to these node numbers:
18-06-12 09:46:49.482 INFO SYNC [p2p-worker-9]: <import-status: node = c33d5a, hash = be4867, number = 405384, txs = 0, result = IMPORTED_BEST, time elapsed = 1 ms>
18-06-12 10:58:36.047 INFO SYNC [p2p-worker-6]: <import-status: node = c33d22, hash = 4c3620, number = 405785, txs = 0, result = IMPORTED_BEST, time elapsed = 0 ms>

One thing that stood out is that the block reported as the chainBest was the first block processed by a p2p-worker-* thread after a batch of blocks processed by sync-ib nodes:

From these numbers I understand that I should not be able to access blocks that have higher numbers than 405384, but fom the Java API I can do that. This suggests that the number reported as chainBestBlock is not acurate.

You can see below that I can get TransactionDetails from a Block that is higher than the current chainBestBlock:
405384 < 405583

L.E: Kernel version is: v0.2.7.1bbeec1

[AionJayT]

fixed in 9551b20

[theotniel]

@AionJayT We integrated the new API jar and used it with the 0.2.8 kernel. While the node was syncing we were receiving correct numbers for chain-best and network best.
However, when the node was effectively synced, we were getting rather weirn numbers:

1. If the node had been running for a while, the chain best is a significantly larger number than the network best:
   

2. If the node was just started, the numbers stay in a range 1-2 from each other but they drift apart after a while, with the chain best becoming larger than the network best:
   

Also, as you can see, the SyncInfo.syncing boolean is always true. Maybe I'm misunderstanding the meaning it has, but I was expecting this to be false when the node is in a synced state (though this is not really critical for us since we don't rely on it's meaning for now)

[AionJayT]

@centrys-otniel this should be the kernel issues. The syncMgr maybe doesn't update the status on-time
I will take a look.

[AionJayT]

fix the issue in the kernel PR.
the kernel binary will publish in the mid of July
aionnetwork/aion#544