CCIP-4447 Promwrapper for OCR3 plugins and factories #15521
Conversation
AER Report: CI Core ran successfully ✅AER Report: Operator UI CI ran successfully ✅ |
mateusz-sekara
requested review from
EasterTheBunny,
winder,
rstout,
asoliman92,
makramkd,
dimkouv,
0xAustinWang and
b-gopalswami
github-merge-queue
bot
removed this pull request from the merge queue due to failed status checks
| result, err := withObservedExecution(p, reports, func() ([]ocr3types.ReportPlus[RI], error) { | ||
| return p.ReportingPlugin.Reports(ctx, seqNr, outcome) | ||
| }) | ||
| p.trackReports(reports, len(result)) |
There was a problem hiding this comment.
Does it mean that we are tracking report even when we have an error?
There was a problem hiding this comment.
reports should be nil or empty in that case
| ) | ||
|
|
||
| var ( | ||
| buckets = []float64{ |
There was a problem hiding this comment.
Are these buckets applicable for all chains? Solana might be significantly faster than Ethereum. I wonder if we should have buckets per chain.
There was a problem hiding this comment.
For CCIP it shouldn't be a big deal. Most of the work we do in OCR is reading from the database (ChainReader), doing some data transformation, reaching 3rd party APIs etc. There is still some communication with the chain involved but it shouldn't be that relevant. However, we might come back to that after seeing different values on non-evm chains
There was a problem hiding this comment.
To answer your question: yes, buckets are applicable for all chains
* PoC * Basic implementation * Basic implementation * Basic implementation * fixes * fixes * fixes * fixes * fixes * fixes * fixes * fixes * fixes * fixes * fixes
Implementing a very basic monitoring layer for OCR3 Reporting Plugins to support similar behavior to what we have for OCR2 https://github.com/smartcontractkit/chainlink/blob/develop/core/services/ocr2/plugins/promwrapper/plugin.go
Compared to the OCR2 implementation some labels like
chainType,configDigestandoracleIDwere removed. This is done to reduce the cardinality of the metrics published to Prometheus. Intentionally, starting with a fundamental set of metrics which can be extended in the future if needed.These metrics will help us answer the following observability questions: