Monerod crashes when synchronizing on Windows.

[Folindril]

When synchronizing the blockchain from 0, most of it is synchronized without problems. Towards the end, monerod starts to crash at random times without any errors. There were cases when the crash occurred in the middle of drawing a log line on the screen. There are also crashes with irreversible damage to the blockchain files. Work in a synchronized state is stable, the bug clearly occurs only when lagging behind the network. This bug occurs either on certain hardware or on a certain OS.This bug is present on all versions 0.18. Did not check earlier versions due to incompatibility of blockchain files.

Version: 0.18.3.4 Windows 64-bit
OS: Windows 11
Processor: Intel Core i5-7600K 3.80GHz
RAM: 16Gb

[selsta]

Do you have any stability issues on your system? Or do you have a different computer where you can test syncing to see if it crashes? There is no known crash bugs in v0.18 that match what you are describing, at least we didn't get other reports about this on Windows.

[Folindril]

Do you have any stability issues on your system? Or do you have a different computer where you can test syncing to see if it crashes? There is no known crash bugs in v0.18 that match what you are describing, at least we didn't get other reports about this on Windows.

No, there are no problems with other applications (mostly games). Monero mining also works stably with a synchronized node.
I have a laptop on Windows 10 in the same network, where synchronization occurs without crashes. No additional parameters in the command line helped me either. Therefore, I decided that this is a bug and it is tied to a certain architecture.

[Folindril]

Now I have run new synchronization again with log detalisation level 1 to detect on which block the first crash will occur. It will take a few days because crashes start towards the end.

[selsta]

The problem is sometimes the last log message is not related to the crash, a stack trace would ideally be needed but that's a bit more complicated on Windows.

[Folindril]

Now I have run new synchronization again with log detalisation level 1 to detect on which block the first crash will occur. It will take a few days because crashes start towards the end.

Done. Block 1650925 (2018-08-31).
Log:

2024-11-17 11:19:56.367 I +++++ BLOCK SUCCESSFULLY ADDED
2024-11-17 11:19:56.368 I id:   <6d920a7d01f688c5f3da5dbb4e7bca6cc67e19b104fc2ef9d56a89b4fad5d9de>
2024-11-17 11:19:56.368 I PoW:  <ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff>
2024-11-17 11:19:56.368 I HEIGHT 1650925, difficulty:   56493714946
2024-11-17 11:19:56.368 I block reward: 3.986338323206(3.969986323206 + 0.016352000000), coinbase_weight: 102, cumulative weight: 53514, 1(0/0)ms
2024-11-17 11:19:56.663 I [91.198.115.225:18380 OUT] 18284 bytes received for category command-1001 initiated by us
2024-11-17 11:19:56.753 I [193.142.4.150:18085 OUT] 2554507 bytes received for category command-2004 initiated by peer
2024-11-17 11:19:56.753 I [193.142.4.150:18085 OUT] Received NOTIFY_RESPONSE_GET_OBJECTS (20 blocks)
2024-11-17 11:19:56.753 I [193.142.4.150:18085 OUT] [0] state: received objects in state synchronizing
2024-11-17 11:19:56.755 I [193.142.4.150:18085 OUT] Failed to lock m_sync_lock, going back to download
2024-11-17 11:19:59.254 I Transaction added to pool: txid <d71ced687cf92480e018b61908911e2383f85b9b39c456678b6fa92457f7b38b> weight: 14927 fee/byte: 159577, count: 1
2024-11-17 11:19:59.255 I [91.198.115.225:18380 OUT] Remote blockchain height: 3283270, id: <985e0b82cdaaa1dd78da8a57a8ad512de6b1407cde839815d4d9f1fa76695d10>
2024-11-17 11:19:59.256 I [91.198.115.225:18380 OUT] [0] state: requesting callback in state synchronizing
2024-11-17 11:19:59.257 I [193.142.4.150:18085 OUT] [0] state: resuming in state synchronizing
2024-11-17 11:19:59.257 I [91.198.115.225:18380 OUT] New connection handshaked, pruning seed 0

[selsta]

There is nothing interesting in the logs unfortunately, the crash happens after the log message is printed. A stack trace would be required to look into this further.