This paves the way for better object lifetime management.
Concretely, it makes it possible to:
- have different stores backing chain and state data.
- having the same datastore library, but using different parameters.
- attach different caching layers/policies to each class of data, e.g.
sizing caches differently.
- specifying different retention policies for chain and state data.
This separation is important because:
- access patterns/frequency of chain and state data are different.
- state is derivable from chain, so one could never expunge the chain
store, and only retain state objects reachable from the last finality
in the state store.
Some components like go-ds-measure, go-ipfs-blockstore and go-bitswap
expose metrics via ipfs/go-metrics-interface, but Lotus never injects
the Prometheus exporter (ipfs/go-metrics-prometheus). Therefore, those
metrics never surface in instrumentation.
Instead, Lotus uses OpenCensus directly.
This commit injects the Prometheus exporter for go-metrics-interface, and
instructs the OpenCensus Prometheus exporter to use the DefaultRegistry.
This has the effect of exposing blending the metrics of both metrics
libraries.
With this patch, the datastore, cache utilisation, and bitswap metrics
are now exported via the /debug/metrics endpoint.
This commit also fixes an issue where the metrics scope was empty, making
go-metrics-interface default to "<no-scope>". Angle brackets are inadmissible
characters for Prometheus, so it was refusing to export the affected metrics.
(These were the ARC cache metrics.)
Motivation:
* Run lotus with the race detector enabled (primary motivation).
* Allow multiple lotus nodes in a process (not a high priority).
Previously, the journal was shared between all lotus instances, but it was
initialized for every new node. This caused safety problems in tests (at a
minimum).
This patch explicitly passes the journal to all services that need it.
