This replaces the builtin byte slice with an interface for the index
reader. This allows the complex decoding of the index file format
to be used against more generalized implementations.
This commit introduces error returns in various places and is explicit
about closing persisted blocks.
{Index,Chunk,Tombstone}Readers are more consistent about their Close()
method. Whenever a reader is retrieved, the corresponding close method
must eventually be called. We use this to track pending readers against
persisted blocks.
Querier's against the DB no longer hold a read lock for their entire
lifecycle. This avoids long running queriers to starve new ones when we
have to acquire a write lock when reloading blocks.
This change loads the full symbol table when we open a persisted block
and allocates a string for each. This ensures that strings retrieved
through the index can be used after the block was closed.
Before we backed the strings by the mmap'd byte regions which would
segfault in this case.
Also remove an inconsistency in the disk format and move both offset
tables to the end (breaking change).
This adds various new locks to replace the single big lock on
the head. All parts now must be COW as they may be held by clients
after initial retrieval.
Series by ID and hashes are now held in a stripe lock to reduce
contention and total holding time during GC. This should reduce
starvation of readers.
Change index persistence for series to not be accumulated in memory
before being written as one large batch. `Labels` and `ChunkMeta`
objects are reused.
This cuts down memory spikes during compaction of multiple blocks
significantly.
As part of the the Index{Reader,Writer} now have an explicit notion of
symbols and series must be inserted in order.
This has been a common source of hard to debug issues. Its a premature
and unbenchmarked optimization and semantically, we want ChunkMetas to
be references in all changed cases.
Fabian Reinartz
Goutham Veeramachaneni