Hi everyone, We use bitcoin core with hundreds of wallets. The only efficient way we have found to do so is :
- wait for any bitcoin-cli command to terminate before issuing another one
- never issue any bitcoin-cli command unless tip update is 1
- always unload a wallet after loading and using it
Now, we observ that bitcoin core systematically crashes every 20-30 minutes, then automatically restarts. And the problem is, it has to re-sync from about 0.997, which easily takes 15-20 minutes. During that time, the daemon cannot be used.
If we don’t load/unload wallets, the daemon is stable and keeps to 1, no issue.
System information Bitcoin Core RPC client version v0.20.0.0-ga62f0ed64f8bbbdfe6467ac5ce92ef5b5222d1bd Compiled and run on Debian 9, on a VM with 16 mo memory. Launched that way : /bitcoind -daemon -walletrbf -conf=/root/.bitcoin/bitcoin.conf -fallbackfee=0.00001 -rpcworkqueue=128 -txindex=1
bitcoin.conf : server=1 dbcache=16384 par=1 blocksonly=1 maxuploadtarget=137 maxconnections=16 rpcuser=bitcoinrpc rpcpassword=xxxxxx rpcallowip=127.0.0.1 testnet=0 txindex=1 walletbroadcast=1 txconfirmtarget=1
Please note that the same behaviour was happening with older releases (from 0.17.x).
Thanks for any help. Please ask for any needed information / testing. Regards,