Celery WorkerLostError Worker 过早退出：信号 6 (SIGABRT)

Question

我想让异步工作同步完成。我使用couchbase中数据的id作为redis的key。我再次使用 redis 的值更新 couchbase。同步运行时可以正常运行，但异步执行时就失败了。

from couchbase.cluster import Cluster
from couchbase.cluster import PasswordAuthenticator
from couchbase.n1ql import N1QLQuery
from couchbase.n1ql import N1QLRequest
from elasticsearch import Elasticsearch
from time import sleep
from urllib import request
from wikidata.client import Client
from openpyxl import Workbook
import base64
import json
import osmapi
import re
import redis
import urllib
import hashlib
import time
import requests
import csv
from functools import reduce

wikid_client = Client()
es_client = Elasticsearch("--")
cluster = Cluster('couchbase://--')
authenticator = PasswordAuthenticator('ATLAS-DEV','ATLAS-DEV')
cluster.authenticate(authenticator)
bucket_test = cluster.open_bucket('ATLAS-DEV')
redis_db = redis.StrictRedis(host='--', port=6379, db=10, decode_responses=True)
map_api = osmapi.OsmApi()
N1QLQuery.timeout = 3600


def deep_get(dictionary, keys, default=None):
    return reduce(lambda d, key: d.get(key, default) if isinstance(d, dict) else default, keys.split("."), dictionary)

query = N1QLQuery("SELECT meta().id, * FROM `ATLAS-DEV` WHERE class_type = 'REGION' AND (codes.osm IS NOT NULL OR codes.osm != '') LIMIT 20")
for k in bucket_test.n1ql_query(query):
    # print("k : ")
    # print(k)
    document_id = k.get("id")
    # print("document_id : " + document_id)
    osm_id = deep_get(k, "ATLAS-DEV.codes.osm")
    polygon = redis_db.hget(name="osm:polygons", key=osm_id)
    if polygon is not None:
        k['ATLAS-DEV'].update({'boundaries': polygon})
        bucket_test.upsert(document_id, k['ATLAS-DEV'])

这是我的同步作品。效果很好。它可以更新沙发底座。

但这是我的芹菜工作代码。

from app.jobs.task import each_k
query = N1QLQuery("SELECT meta().id, * FROM `ATLAS-DEV` WHERE class_type = 'REGION' AND (codes.osm IS NOT NULL OR codes.osm != '') LIMIT 5")
for k in bucket_test.n1ql_query(query):
    # print("k : ")
    # print(k)
    each_k.delay(k)




# /app/jobs/task.py

def deep_get(dictionary, keys, default=None):
    return reduce(lambda d, key: d.get(key, default) if isinstance(d, dict) else default, keys.split("."), dictionary)

@app.task
@provide_redis
def each_k(redis_db, k):
    print("레디스 디비 : ")
    print(redis_db)
    document_id = k.get("id")
    print("document_id : " + document_id)
    osm_id = deep_get(k, "ATLAS-DEV.codes.osm")
    polygon = redis_db.hget(name="osm:polygons", key=osm_id)
    print("폴리곤 : ")
    print(polygon)
    if polygon is not None:
        k['ATLAS-DEV'].update({'boundaries': polygon})
        bucket_test.upsert(document_id, k['ATLAS-DEV'])

它回来了

[warn] kevent: Bad file descriptor
python-couchbase: self->nremaining == 0 at src/oputil.c:67. Abort[2019-06-26 15:13:03,675: ERROR/MainProcess] Task handler raised error: WorkerLostError('Worker exited prematurely: signal 6 (SIGABRT).')
Traceback (most recent call last):
  File "/Users/n18016/Library/Python/3.7/lib/python/site-packages/billiard/pool.py", line 1226, in mark_as_worker_lost
    human_status(exitcode)),
billiard.exceptions.WorkerLostError: Worker exited prematurely: signal 6 (SIGABRT).
[2019-06-26 15:13:03,701: ERROR/MainProcess] Process 'ForkPoolWorker-11' pid:7697 exited with 'signal 6 (SIGABRT)'
[2019-06-26 15:13:03,702: ERROR/MainProcess] Process 'ForkPoolWorker-10' pid:7696 exited with 'signal 6 (SIGABRT)'
[2019-06-26 15:13:03,703: ERROR/MainProcess] Task handler raised error: WorkerLostError('Worker exited prematurely: signal 6 (SIGABRT).')
Traceback (most recent call last):
  File "/Users/n18016/Library/Python/3.7/lib/python/site-packages/billiard/pool.py", line 1226, in mark_as_worker_lost
    human_status(exitcode)),
billiard.exceptions.WorkerLostError: Worker exited prematurely: signal 6 (SIGABRT).
[2019-06-26 15:13:03,704: ERROR/MainProcess] Task handler raised error: WorkerLostError('Worker exited prematurely: signal 6 (SIGABRT).')
Traceback (most recent call last):
  File "/Users/n18016/Library/Python/3.7/lib/python/site-packages/billiard/pool.py", line 1226, in mark_as_worker_lost
    human_status(exitcode)),
billiard.exceptions.WorkerLostError: Worker exited prematurely: signal 6 (SIGABRT).

Answer 1

https://github.com/celery/celery/issues/4113

celery worker --pool solo ...

这个 cli 选项救了我的命。效果很好。这个命令可以很好地运行异步单线程。

这可能不是所有情况的答案。如果您想在

single thread

中完成此操作，这就是答案。