Detect when GETBLOCKTXN requests are ignored. #22307

pull rebroad wants to merge 1 commits into bitcoin:master from rebroad:DetectIngoredGetblocktxns changing 1 files +23 −3
  1. rebroad commented at 8:07 AM on June 22, 2021: contributor

    Fixes issue #22295

    (The next thing to do is to make it so that the PartialBlock can be reused, rather than it requesting a full block).

  2. Detect when GETBLOCKTXN requests are ignored.
    (Fixes #22295)
    d9244f1e8d
  3. rebroad force-pushed on Jun 22, 2021
  4. fanquake commented at 9:52 AM on June 22, 2021: member

    This doesn't compile:

    net_processing.cpp:3540:33: error: calling function 'State' requires holding mutex 'cs_main' exclusively [-Werror,-Wthread-safety-analysis]
            CNodeState *nodestate = State(pfrom.GetId());
                                    ^
    1 error generated.
    
  5. DrahtBot added the label P2P on Jun 22, 2021
  6. DrahtBot commented at 3:05 PM on June 22, 2021: contributor

    <!--e57a25ab6845829454e8d69fc972939a-->

    The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

    <!--174a7506f384e20aa4161008e828411d-->

    Conflicts

    Reviewers, this pull request conflicts with the following ones:

    If you consider this pull request important, please also help to review the conflicting pull requests. Ideally, start with the one that should be merged first.

  7. DrahtBot cross-referenced this on Jun 22, 2021 from issue refactor: Release cs_main before MaybeSendFeefilter by MarcoFalke
  8. jnewbery cross-referenced this on Jun 22, 2021 from issue ignored getblocktxn takes longer to detect than is possible by rebroad
  9. jnewbery commented at 6:41 PM on June 22, 2021: member

    I'm not sure this is the direction we want to be moving in. Ideally, our transaction and block relay logic are mostly separate. It'd be good to disentangle any logic/state in our net_processing code that has tight coupling between the two.

    The root cause here seems to be that the block download timeout is way too long in most cases, and should be significantly reduced once the node is out of initial sync and at the tip. That timeout isn't specific to compact blocks and could be reduced for all block downloads.

  10. fanquake commented at 2:07 AM on June 28, 2021: member

    Going to close. Seems like it would be better to continue discussion on #22295 for now. Compilation issues also still haven't been addressed.

  11. fanquake closed this on Jun 28, 2021

  12. bitcoin locked this on Aug 16, 2022

github-metadata-mirror

This is a metadata mirror of the GitHub repository bitcoin/bitcoin. This site is not affiliated with GitHub. Content is generated from a GitHub metadata backup.
generated: 2026-05-19 06:53 UTC