pull in geth changes for state recreation #2005
Conversation
cla-bot
bot
added
the
s
magicxyyz
changed the title
This reverts commit 8eebd46.
|
|
||
| blockCountRequiredToFlushDirties := builder.execConfig.Caching.BlockCount | ||
| makeSomeTransfers(t, ctx, builder, blockCountRequiredToFlushDirties) | ||
|
|
There was a problem hiding this comment.
can we add here a check that StateAndHeaderByNumber will now fail?
Or maybe force state to drop after the end of current test - and check that StateAndHeaderByNumber fails then?
There was a problem hiding this comment.
addressed in #2442
| } | ||
| } | ||
|
|
||
| func TestGettingStateForRPCHybridArchiveNode(t *testing.T) { |
There was a problem hiding this comment.
Is the logic here different from above other than config? can we merge the functions?
There was a problem hiding this comment.
addressed in #2442
| var wgCallers sync.WaitGroup | ||
| for j := 0; j < threads && ctx.Err() == nil; j++ { | ||
| wgCallers.Add(1) | ||
| go func() { |
There was a problem hiding this comment.
this gofunc is quite weird and deserves at least some explanation in comments
There was a problem hiding this comment.
addressed in #2442
| }() | ||
| api := builder.L2.ExecNode.Backend.APIBackend() | ||
| db := builder.L2.ExecNode.Backend.ChainDb() | ||
| i := 1 |
There was a problem hiding this comment.
this is a sync variable between multiple threads.. the name "i" doesn't really fit it
There was a problem hiding this comment.
addressed in #2442
| prefix = append([]byte("b"), prefix...) | ||
| it := db.NewIterator(prefix, nil) | ||
| defer it.Release() | ||
| if it.Next() { |
There was a problem hiding this comment.
suggestion: to speed up polling and hit more edge-cases, each thread could cache blockhashes that it already got a non-error response for in previous loops and skip them the next time iterator gets them
There was a problem hiding this comment.
alternatively, we could poll old blockhashes and make sure they don't return the "ahead of current block" error.
There was a problem hiding this comment.
this test is mostly intended to be a regression test for a specific issue, I added a comment describing the details in the follow up PR: #2442
there should be only one key with the prefix (unless there was a reorg) so there is no need of caching. If the entry is found, the recentBlock (previously i) is advanced to the next block, and the iterator is created for the new prefix.
Let me know if you think that extending this test makes sense to cover more cases then the targeted issue
StateAndHeaderByNumberOrHash