Tendermint Block Pruning

[alexanderbez]

Description

Implement Tendermint block pruning support via ResponseCommit.RetainHeight:

• Introduce new min-retain-blocks in the app.toml configuration with an equivalent CLI flag
• Call GetBlockRentionHeight during ABCI Commit which follows https://github.com/tendermint/spec/blob/master/rfc/001-block-retention.md

closes: #6164

/cc @marbar3778 @erikgrinaker

---

Before we can merge this PR, please make sure that all the following items have been
checked off. If any of the checklist items are not applicable, please leave them but
write a little note why.

• Targeted PR against correct branch (see CONTRIBUTING.md)
• Linked to Github issue with discussion and accepted design OR link to spec that describes this work.
• Code follows the module structure standards.
• Wrote unit and integration tests
• Updated relevant documentation (docs/) or specification (x/<module>/spec/)
• Added relevant godoc comments.
• Added a relevant changelog entry to the Unreleased section in CHANGELOG.md
• Re-reviewed Files changed in the Github PR explorer
• Review Codecov Report in the comment section below once CI passes

[codecov]

Codecov Report

Merging #7265 into master will increase coverage by 0.65%.
The diff coverage is 51.33%.

@@            Coverage Diff             @@
##           master    #7265      +/-   ##
==========================================
+ Coverage   55.60%   56.25%   +0.65%     
==========================================
  Files         457      584     +127     
  Lines       27440    40563   +13123     
==========================================
+ Hits        15257    22820    +7563     
- Misses      11083    15823    +4740     
- Partials     1100     1920     +820     

[tac0turtle]

utACK

[fedekunze]

LGTM pending comments for maintainers

[fedekunze]

Unblocking

[sahith-narahari]

lgtm

[erikgrinaker]

I think this should use KeepEvery, not Interval. Consider the case of KeepEvery: 1000 and Interval: 10. At height 10890, this will allow block pruning up to 10880, even though the last state height persisted to disk is 10000 - if the node restarts, it will try to replay from 10000, but can't since the blocks are gone.

[alexanderbez]

You're absolutely right! Can't believe I missed this. Interval is when the heights are actually pruned from disk -- nomenclature always confuses me...

[alexanderbez]

However, in this case H=10890, and X=H-KeepEvery=9890 which isn't totally correct either. I think we need to do X=H-(H%KeepEvery)?

[erikgrinaker]

I think we need to do X=H-(H%KeepEvery)?

Yes, that's more accurate. H-KeepEvery would work, but has an excessive safety margin. If you go for minNonZero then take care when H<KeepEvery, since this expression results in 0 and thus would be ignored.

[alexanderbez]

take care when H<KeepEvery, since this expression results in 0 and thus would be ignored.

Correct, in this case, we must be sure to return 0 as to not prune anything because that would be bad!

[alexanderbez]

Updated. @erikgrinaker could you take another look?

[erikgrinaker]

Looks good!

It occurred to me that another way to handle this would be to simply return a configuration error if minRetainBlocks was set lower than all these other parameters - instead of magically extending it. Not really sure which UX is better.