This is based on my experience while debugging https://github.com/prometheus/prometheus/issues/3943.
I needed to deduct few things, and all that would be just bit easier with these two logs:
- new block's ULID on each compaction.
- actual list of Blocks (ulid + time range) on Prometheus startup (easy to log that while repairing blocks).
We don't really need blocks that takes part in compaction - that can be deducted easily based on time ranges of blocks we have currently in system.
What do you think?
Signed-off-by: Bartek Plotka <bwplotka@gmail.com>
Aligning postings list for similar keys close to each other improves
page cache hit rates in typical queries that select postings for
multiple label pairs with the same name.
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.
IDs for new series are handed out before the postings are locked. Thus
series are not indexed in order of their IDs, which could result in only
partially sorted postings list.
Iterating over those silently skipped elements as the sort invariant was
violated.
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.
This changes the structure to a single WAL backed by a single head
block.
Parts of the head block can be compacted. This relieves us from any head
amangement and greatly simplifies any consistency and isolation concerns
by just having a single head.
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.
Bartek Plotka
Fabian Reinartz
Goutham Veeramachaneni
Simon Pasquier
Shubheksha Jalan
ranbochen