2022-09-07 06:20:30 +00:00
|
|
|
import asyncio
|
2022-09-20 22:03:02 +00:00
|
|
|
from os import getenv
|
2022-09-07 06:20:30 +00:00
|
|
|
|
2022-09-07 06:20:30 +00:00
|
|
|
import db
|
|
|
|
import util
|
2022-09-13 21:17:46 +00:00
|
|
|
from processing import process
|
|
|
|
|
2022-09-30 06:22:22 +00:00
|
|
|
KEYNAME = "queue"
|
2022-09-16 16:09:49 +00:00
|
|
|
|
2022-09-30 06:22:22 +00:00
|
|
|
# Chunk size
|
2022-09-20 17:13:46 +00:00
|
|
|
CHUNK_SIZE = int(getenv("MONOLITH_INGEST_CHUNK_SIZE", "900"))
|
|
|
|
ITER_DELAY = float(getenv("MONOLITH_INGEST_ITER_DELAY", "0.5"))
|
2022-09-07 06:20:30 +00:00
|
|
|
|
2025-01-23 11:29:48 +00:00
|
|
|
INGEST_INCREASE_BELOW = int(getenv("MONOLITH_INGEST_INCREASE_BELOW", "2500"))
|
|
|
|
INGEST_DECREASE_ABOVE = int(getenv("MONOLITH_INGEST_DECREASE_ABOVE", "10000"))
|
|
|
|
|
|
|
|
INGEST_INCREASE_BY = int(getenv("MONOLITH_INGEST_INCREASE_BY", "100"))
|
|
|
|
INGEST_DECREASE_BY = int(getenv("MONOLITH_INGEST_DECREASE_BY", "100"))
|
2022-09-16 16:09:49 +00:00
|
|
|
log = util.get_logger("ingest")
|
|
|
|
|
2025-01-23 11:29:48 +00:00
|
|
|
INGEST_MAX = int(getenv("MONOLITH_INGEST_MAX", "1000000"))
|
|
|
|
INGEST_MIN = int(getenv("MONOLITH_INGEST_MIN", "100"))
|
2022-09-07 06:20:30 +00:00
|
|
|
|
2022-09-07 06:20:30 +00:00
|
|
|
class Ingest(object):
|
|
|
|
def __init__(self):
|
|
|
|
name = self.__class__.__name__
|
|
|
|
self.log = util.get_logger(name)
|
2022-10-21 06:20:30 +00:00
|
|
|
self.current_chunk = 0
|
2022-09-23 07:32:29 +00:00
|
|
|
self.log.info(
|
|
|
|
(
|
|
|
|
"Starting ingest handler for chunk size of "
|
|
|
|
f"{CHUNK_SIZE} every {ITER_DELAY} seconds."
|
|
|
|
)
|
|
|
|
)
|
2022-09-07 06:20:30 +00:00
|
|
|
|
|
|
|
async def run(self):
|
|
|
|
while True:
|
2022-09-13 21:17:46 +00:00
|
|
|
await self.get_chunk()
|
2022-10-21 06:20:30 +00:00
|
|
|
self.log.debug(f"Ingest chunk {self.current_chunk} complete")
|
|
|
|
self.current_chunk += 1
|
2022-09-07 06:20:30 +00:00
|
|
|
await asyncio.sleep(ITER_DELAY)
|
|
|
|
|
2022-09-13 21:17:46 +00:00
|
|
|
async def get_chunk(self):
|
2025-01-23 11:29:48 +00:00
|
|
|
global CHUNK_SIZE
|
2022-10-21 06:20:30 +00:00
|
|
|
length = await db.ar.llen(KEYNAME)
|
2022-10-21 06:20:30 +00:00
|
|
|
if length > CHUNK_SIZE:
|
|
|
|
length = CHUNK_SIZE
|
|
|
|
if not length:
|
2022-09-30 06:22:22 +00:00
|
|
|
return
|
2025-01-23 11:29:48 +00:00
|
|
|
ingested = await process.spawn_processing_threads(self.current_chunk, length)
|
|
|
|
|
|
|
|
if ingested < INGEST_INCREASE_BELOW:
|
|
|
|
if CHUNK_SIZE + INGEST_INCREASE_BY < INGEST_MAX:
|
|
|
|
self.log.info(
|
|
|
|
(
|
|
|
|
f"Increasing chunk size to "
|
|
|
|
f"{CHUNK_SIZE + INGEST_INCREASE_BY} "
|
|
|
|
f"due to low ingestion ({ingested})"
|
|
|
|
)
|
|
|
|
)
|
|
|
|
CHUNK_SIZE += INGEST_INCREASE_BY
|
|
|
|
else:
|
|
|
|
log.info(f"Chunk size ({CHUNK_SIZE}) at maximum, not increasing above: {INGEST_MAX}")
|
|
|
|
|
|
|
|
elif ingested > INGEST_DECREASE_ABOVE:
|
|
|
|
if CHUNK_SIZE - INGEST_DECREASE_BY > INGEST_MIN:
|
|
|
|
self.log.info(
|
|
|
|
(
|
|
|
|
f"Decreasing chunk size to "
|
|
|
|
f"{CHUNK_SIZE - INGEST_DECREASE_BY}"
|
|
|
|
f"due to high ingestion ({ingested})"
|
|
|
|
)
|
|
|
|
)
|
|
|
|
CHUNK_SIZE -= INGEST_DECREASE_BY
|
|
|
|
else:
|
|
|
|
log.info(f"Chunk size ({CHUNK_SIZE}) at minimum, not decreasing below: {INGEST_MIN}")
|