Fields

• closeLeiosDbHandle ∷ m ()

  Close the database: flush what is in flight, stop whatever it runs behind the scenes, release its connections. Readers and writers handed out earlier are not usable afterwards.

• openReader ∷ HasCallStack ⇒ m (LeiosDbReader m)

  Get a new reader. No interaction between readers or writers.

• openWriter ∷ HasCallStack ⇒ m (LeiosDbWriter m)

  Get a new writer. All writes of all writers are serialised.

• subscribeEbNotifications ∷ HasCallStack ⇒ m (StrictTChan m LeiosEbNotification)

  New EBs and EB closures as they are stored, from the moment of subscription. TODO: make return type more descriptive (e.g. Subscription { getNext :: STM m LeiosEbNotification })

• leiosDbGarbageCollect ∷ HasCallStack ⇒ SlotNo → m ()

  Trigger garbage collection of the LeiosDB.

  This function does not do have to do the actual GC, but rather act as the MARK phase of GC and "create work" for a background SWEEP thread.

  See sqlGarbageCollect for the SQL backend implementation.

• leiosDbPromoteToImmutable ∷ HasCallStack ⇒ [LeiosPoint] → m ()

  Promote the given EBs' bodies and tx closures from volatile into immutable LeiosDb. Takes a batch: the caller holds the ImmutableDB write lock and blocks on the shared writer queue, so this must be one round-trip per copyToImmutableDB pass, not one per certified EB.

• leiosDbSampleStats ∷ HasCallStack ⇒ m LeiosDbStats

  Sample LeiosDbStats counters.

Fields

• volatileEbs ∷ !Int
   
• immutableEbs ∷ !Int
   
• walBytes ∷ !Integer
   

Fields

• LeiosPoint

  the EB's point

• BytesSize

  its size in bytes

Fields

• LeiosPoint

  the EB's point

Fields

• closeReader ∷ m ()
   
• lookupEbBody ∷ HasCallStack ⇒ EbHash → m [(TxHash, BytesSize)]

  The EB "body": tx hashes and sizes in order, no tx bytes.

• lookupEbClosure ∷ HasCallStack ⇒ EbHash → m (Maybe [(TxHash, ByteString)])

  The EB "closure": tx hashes and their bytes, or Nothing if the EB is not complete.

• batchRetrieveTxs ∷ HasCallStack ⇒ EbHash → [Int] → m [(Int, TxHash, Maybe ByteString)]

  Tx bytes for a batch of offsets into one EB.

• testScanEbPoints ∷ HasCallStack ⇒ m [(SlotNo, EbHash)]

  Every announced EB point.

  Not used in production, only used in tests to observe point writes and truncation.

• scanCompleteEbClosuresNotOlderThanSlot ∷ HasCallStack ⇒ SlotNo → m [LeiosPoint]

  EBs whose closure is complete and whose announcer is no older than the given slot. Seeds the ChainDB's acquired-closures set at startup.

Fields

• closeWriter ∷ m ()

  Close writer and flush all remaining writes.

• writeEbPoint ∷ HasCallStack ⇒ LeiosPoint → BytesSize → m (Promise m ())

  Record an announced EB's point and expected size.

• writeEbBody ∷ HasCallStack ⇒ LeiosPoint → LeiosEb → m (Promise m CompletedEbs)

  Persist an EB body. Returns any EBs whose closure this completed.

  The returned EB is only used for tracing and too broad: it can actually only be this same EB which got completed.

• writeTxs ∷ HasCallStack ⇒ [(TxHash, ByteString)] → m (Promise m CompletedEbs)

  Persist tx bodies. Returns the EBs whose closure this completed.

Fields

• await ∷ m a