Can't sync both on Windows and Linux #36428

pbies commented at 4:00 PM on October 3, 2026: none

Is there an existing issue for this?

I have searched the existing issues

Current behaviour

Windows 10 up-to-date and Linux Mint 22.3 up-to-date (desktop, Intel NUC). Two versions tested - 29.4 and 31.1, both GUI (qt) and CLI (bitcoind), both x64.

There is no way to sync the Core with blockchain. There always happen blocks files folder problem, or chainstate problem, or indexes/txindex problem. I am encountering this for past 3 weeks and cannot make one version of Core synced in any way.

There is no specific point in time/blockchain that the errors occur. Memory and disk were tested for errors - nothing. SMART says all is "ok".

Added debug=leveldb but no more info than without it.

Errors appear in random moments when downloading blockchain. Tried also to remove "blocks", "chainstate" and "indexes" folders but it didn't help.

Problems says checksum does not match, for files in all three folders.

Expected behaviour

There should be a way to correctly sync Core to blockchain.

Steps to reproduce

Just run bitcoin-qt or bitcoind and wait for blockchain sync, tried also to temove three folders with data (blocks, chainstate, indexes) and change xor.dat to 0x00000000.

Relevant log output

2026-10-03T15:30:08Z Fatal LevelDB error: Corruption: block checksum mismatch: D:\Bitcoin\chainstate/005625.ldb 2026-10-03T15:30:08Z You can use -debug=leveldb to get more complete diagnostic messages 2026-10-03T15:30:08Z [error] A fatal internal error occurred, see debug.log for details: System error while flushing: Fatal LevelDB error: Corruption: block checksum mismatch: D:\Bitcoin\chainstate/005625.ldb 2026-10-03T15:30:15Z [error] ProcessNewBlock: ActivateBestChain failed (System error while flushing: Fatal LevelDB error: Corruption: block checksum mismatch: D:\Bitcoin\chainstate/005625.ldb) 2026-10-03T15:30:15Z UpdateTip: new best=000000000000000000101d817c2f64840a303df95c778d3daa754b468d928aeb height=536269 version=0x20000000 log2_work=89.430340 tx=334093778 date='2018-08-11T12:27:26Z' progress=0.239833 cache=1292.3MiB(10464526txo) 2026-10-03T15:30:15Z [warning] Flushing large (1 GiB) UTXO set to disk, it may take several minutes 2026-10-03T15:30:15Z [coindb:error] The coins database detected an inconsistent state, likely due to a previous crash or shutdown. You will need to restart bitcoind with the -reindex-chainstate or -reindex configuration option.

How did you obtain Bitcoin Core

Pre-built binaries

What version of Bitcoin Core are you using?

29.4 and 31.1

Operating system and version

Microsoft Windows [Version 10.0.19045.7725] and Linux Mint 22.3 cinnamon

Machine specifications

CPU AMD Ryzen 9 9950X mobo ASRock X670E Taichi PSU be quiet! Dark Power 13 1000W (BN335) RAM Corsair Dominator Platinum RGB schwarz DIMM Kit 64GB, DDR5-6400, CL32-40-40-84, on-die ECC GPU GIGABYTE AORUS GeForce RTX 4090 Xtreme Waterforce 24G, 24GB GDDR6X, HDMI, 3x DP disks NVMe and SSD

pbies commented at 4:01 PM on October 3, 2026: none

2026-10-03T16:01:13Z LevelDB read failure: Corruption: block checksum mismatch: D:\Bitcoin\chainstate/006442.ldb 2026-10-03T16:01:13Z Fatal LevelDB error: Corruption: block checksum mismatch: D:\Bitcoin\chainstate/006442.ldb 2026-10-03T16:01:13Z You can use -debug=leveldb to get more complete diagnostic messages 2026-10-03T16:01:13Z [error] Error reading from database, shutting down.

pbies commented at 4:05 PM on October 3, 2026: none

Both -reindex-chainstate and -reindex does not help. There always will be an error. Deleting the data folders also do not help.

pbies commented at 6:12 PM on October 3, 2026: none

`2026-10-03T18:11:50Z Fatal LevelDB error: Corruption: block checksum mismatch: D:\Bitcoin\indexes\txindex/020871.ldb 2026-10-03T18:11:50Z You can use -debug=leveldb to get more complete diagnostic messages 2026-10-03T18:11:50Z

EXCEPTION: 15dbwrapper_error Fatal LevelDB error: Corruption: block checksum mismatch: D:\Bitcoin\indexes\txindex/020871.ldb D:\Bitcoin Core\bin\bitcoin-qt.exe in txindex `

pbies commented at 7:31 PM on October 3, 2026: none

Restarting Core helps for next few thousands of blocks. But in the end chainstate fails and doing -reindex-chainstate makes the work again from start. It goes till next error and then need to run Core without params so it could go further. Then it will encounter next errors or the same one in chainstate.

pbies commented at 8:55 PM on October 3, 2026: none

2026-10-03T20:55:06Z [error] ReadBlock: Deserialize or I/O error - ReadCompactSize(): size too large: iostream error at FlatFilePos(nFile=2788, nPos=72879842) 2026-10-03T20:55:06Z [error] A fatal internal error occurred, see debug.log for details: Failed to read block. Error: A fatal internal error occurred, see debug.log for details: Failed to read block. 2026-10-03T20:55:06Z Shutdown: In progress... 2026-10-03T20:55:06Z addcon thread exit 2026-10-03T20:55:06Z [error] A fatal internal error occurred, see debug.log for details: Failed to connect best block (Failed to read block.). E2rror: 026-10-03T20:55:06Z opencon thread exit A fatal internal error occurred, see debug.log for details: Failed to connect best block (Failed to read block.).

pbies commented at 9:40 PM on October 3, 2026: none

I think -reindex and -reindex-chainstate both should have been given number of blocks to start at. Other way all the blockchain will be rescanned when it is all ok till some block.

pinheadmz commented at 9:45 PM on October 3, 2026: member

could you start bitcoind with a totally empty data directory on the linux system, using the latest release and stop after the first error (dont delete any files) and report from there?

pbies commented at 9:48 PM on October 3, 2026: none

Yes, already done that few times in last 3 weeks. First error comes from chainstate or indexes/txindex. Then I do chainstate reindex, which helps till one of block files fail. Then do whole -reindex and never come to an end (some files always fail). Error messages are the same, nothing new, nothing extra over what I already shared. Currently it is impossible to rescan blockchain till today in any way without code change.

pinheadmz commented at 9:51 PM on October 3, 2026: member

You shared a lot of logs, and youre using two platforms and two versions and deleting xor.dat etc. Lets just take it one step at a time. Fresh directory, linux, no GUI, whats the first error

pbies commented at 10:28 PM on October 3, 2026: none

First error with empty folder:

2026-10-03T22:24:18Z [error] A fatal internal error occurred, see debug.log for details: Corrupt block found indicating potential hardware failure. 2026-10-03T22:24:18Z [error] ConnectTip: ConnectBlock 0000000000000001678f01e2b7f55253ecbacd403d78e35429644edda01531fb failed, bad-txnmrklroot, hashMerkleRoot mismatch 2026-10-03T22:24:18Z mapport thread exit 2026-10-03T22:24:18Z opencon thread exit 2026-10-03T22:24:18Z addcon thread exit 2026-10-03T22:24:18Z Shutdown: In progress... 2026-10-03T22:24:18Z [error] A fatal internal error occurred, see debug.log for details: Failed to connect best block (bad-txnmrklroot, hashMerkleRoot mismatch). 2026-10-03T22:24:18Z msghand thread exit 2026-10-03T22:24:18Z txindex thread start 2026-10-03T22:24:18Z txindex is enabled at height 274240 2026-10-03T22:24:18Z txindex thread exit 2026-10-03T22:24:18Z Loading 0 mempool transactions from file... 2026-10-03T22:24:18Z Imported mempool transactions from file: 0 succeeded, 0 failed, 0 expired, 0 already there, 0 waiting for initial broadcast 2026-10-03T22:24:18Z initload thread exit 2026-10-03T22:24:18Z net thread exit 2026-10-03T22:24:18Z scheduler thread exit 2026-10-03T22:24:18Z Flushed fee estimates to fee_estimates.dat. 2026-10-03T22:24:18Z [warning] Flushing large (1 GiB) UTXO set to disk, it may take several minutes 2026-10-03T22:24:24Z Shutdown: done

bitcoin.conf:

addnode=37-229-85-197.broadband.kyivstar.net addnode=77-173-132-140.fixed.kpn.net addnode=91-206-17-195.static.deltatelesystems.ru addnode=btc.dohmen.net addnode=dsde833.fornex.org addnode=host93-234-110-95.serverdedicati.aruba.it connect=nuc:8831 addnode=smtp2.igmail.net addnode=unnum-78-27-139-13.domashka.kiev.ua addresstype=legacy changetype=legacy datadir = d:\Bitcoin dbcache = 512 listen = 1 listenonion=0 logips = 1 maxconnections=20 par = 6 peertimeout = 5 rest = 1 rpcpassword = rpcport = 8332 rpcuser = server = 1 txindex = 1 blocksxor=0 includeconf = D:\Bitcoin\bitcoin.conf allowignoredconf=1

Memory tested for 2 passes both with PassMark and standard MemTest86+ - no problems. Drive SMART params all ok.

Please ask for all info in the next comment as I will be unable to respond later. Thank you for understanding.

maflcko added the label Data corruption on Oct 4, 2026

maflcko commented at 9:10 AM on October 4, 2026: member

Memory tested for 2 passes both with PassMark and standard MemTest86+ - no problems. Drive SMART params all ok.

It is possible that the hardware corruption only shows when there is enough load to heat the devices for corruption to happen. Also, not sure if your storage is external, but USB-attached drives are frequently faulty.

However, everything you shared so far points at a data corruption due to faulty hardware.

pbies commented at 1:00 PM on October 4, 2026: none

Yep, that was RAM failure. I lowered down the frqs and errors are rarer.

maflcko commented at 1:47 PM on October 4, 2026: member

Yep, that was RAM failure.

Ok, closing for now

maflcko closed this on Oct 4, 2026