Wallet tx not reaching the network after peer set turns over; automatic rebroadcast doesn't recover it #54

Closed
opened 2026-11-14 21:48:33 +00:00 by MWOWM · 2 comments
Member

Version: dobbscoin-qt 0.14.1 (protocol 70005), Windows 10. Peers were on /EndTimes:0.14.0/.

What happens

A sent transaction stayed unconfirmed for about 40 minutes. During that time the dominant pool (1PnkhWyMG1eWQ3Y3WRV1X5txZxNt9wL2Sg) mined 15 blocks, including some with other people's lower-fee transactions, which suggests the pool's node never received ours. A manual sendrawtransaction got it confirmed in the next block.

Timeline (UTC, 2026-11-14)

  • 20:26:58: Relaying wtx ce91a7ba… (initial broadcast)
  • around 21:00: all outbound peers drop and reconnect; socket send error An operation was attempted on something that is not a socket. (10038) and socket recv error An established connection was aborted by the software in your host machine. (10053)
  • 21:00:16: Relaying wtx ce91a7ba… (automatic rebroadcast, alongside the socket errors)
  • 21:02: pool mines block 1948310 with another tx (about 0.00001/kB) but not ours (about 0.00006/kB)
  • 21:07: manual sendrawtransaction
  • 21:08: confirmed in block 1948315

Tx: ce91a7ba322b848d493a2436c6556745565ace40a925a07c35f43dd194cb64a9. All 23 inputs were confirmed and unspent; the fee was well above relayfee (0.00001).

Status

Unconfirmed root cause. The 21:00 rebroadcast may have gone to sockets that were closing. Another possibility is that the 0.14.0 peers rejected the tx and only accepted it later. Running with debug=net would show which.

Possible fixes

  • Rebroadcast wallet txs to newly connected peers.
  • Retry soon after a rebroadcast that overlapped socket errors.

Workaround

dobbscoin-cli sendrawtransaction $(dobbscoin-cli getrawtransaction <txid>)

**Version:** dobbscoin-qt 0.14.1 (protocol 70005), Windows 10. Peers were on `/EndTimes:0.14.0/`. ### What happens A sent transaction stayed unconfirmed for about 40 minutes. During that time the dominant pool (`1PnkhWyMG1eWQ3Y3WRV1X5txZxNt9wL2Sg`) mined 15 blocks, including some with other people's lower-fee transactions, which suggests the pool's node never received ours. A manual `sendrawtransaction` got it confirmed in the next block. ### Timeline (UTC, 2026-11-14) - 20:26:58: `Relaying wtx ce91a7ba…` (initial broadcast) - around 21:00: all outbound peers drop and reconnect; `socket send error An operation was attempted on something that is not a socket. (10038)` and `socket recv error An established connection was aborted by the software in your host machine. (10053)` - 21:00:16: `Relaying wtx ce91a7ba…` (automatic rebroadcast, alongside the socket errors) - 21:02: pool mines block 1948310 with another tx (about 0.00001/kB) but not ours (about 0.00006/kB) - 21:07: manual `sendrawtransaction` - 21:08: confirmed in block 1948315 Tx: `ce91a7ba322b848d493a2436c6556745565ace40a925a07c35f43dd194cb64a9`. All 23 inputs were confirmed and unspent; the fee was well above `relayfee` (0.00001). ### Status Unconfirmed root cause. The 21:00 rebroadcast may have gone to sockets that were closing. Another possibility is that the 0.14.0 peers rejected the tx and only accepted it later. Running with `debug=net` would show which. ### Possible fixes - Rebroadcast wallet txs to newly connected peers. - Retry soon after a rebroadcast that overlapped socket errors. ### Workaround `dobbscoin-cli sendrawtransaction $(dobbscoin-cli getrawtransaction <txid>)`
Author
Member

Fix written for EndTimes 0.21.2, going into beta 2. Bitcoin Core 0.21 re-sends unconfirmed wallet transactions only every 12 to 36 hours (0.14 did every 0 to 30 minutes); EndTimes will re-send every 5 to 15 minutes, for unconfirmed, unconflicted wallet transactions more than 5 minutes older than the last block. Core 0.21's other retry (the unbroadcast set) stops as soon as any one peer requests the transaction, which is exactly the case here: the peers that took it dropped. #53 turned out to be this same issue seen from the pool's side: the pool node's templates stayed coinbase-only until the manual re-send. I'll close this when beta 2 ships.

Fix written for EndTimes 0.21.2, going into beta 2. Bitcoin Core 0.21 re-sends unconfirmed wallet transactions only every 12 to 36 hours (0.14 did every 0 to 30 minutes); EndTimes will re-send every 5 to 15 minutes, for unconfirmed, unconflicted wallet transactions more than 5 minutes older than the last block. Core 0.21's other retry (the unbroadcast set) stops as soon as any one peer requests the transaction, which is exactly the case here: the peers that took it dropped. #53 turned out to be this same issue seen from the pool's side: the pool node's templates stayed coinbase-only until the manual re-send. I'll close this when beta 2 ships.
Author
Member

The fix (wallet re-sends unconfirmed transactions every 5 to 15 minutes instead of every 12 to 36 hours) shipped in EndTimes 0.21.2 beta 2 and is in beta 3: https://git.subgenius.finance/SubGeniusFinance/dobbscoin-source/releases/tag/v0.21.2-beta3

Closing as promised. Reopen if a send still sits unconfirmed on beta 3.

The fix (wallet re-sends unconfirmed transactions every 5 to 15 minutes instead of every 12 to 36 hours) shipped in EndTimes 0.21.2 beta 2 and is in beta 3: https://git.subgenius.finance/SubGeniusFinance/dobbscoin-source/releases/tag/v0.21.2-beta3 Closing as promised. Reopen if a send still sits unconfirmed on beta 3.
MWOWM closed this issue 2026-11-21 14:50:31 +00:00
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
SubGeniusFinance/dobbscoin-source#54
No description provided.