How to handle an array of `4`s returned by redis-py `pipeline.execute()`?

Question

I'm using redis-py to process bulk insertions into a Redis store.

I wrote the following very simple method:

import redis

def push_metadata_to_redis(list_of_nested_dictionaries):

    redis_client = redis.Redis(host='localhost', port=6379, db=0)
    redis_pipeline = redis_client.pipeline(transaction=False)

    for dictionary in list_of_nested_dictionaries:
        for k, inner_dict in dictionary.items()
            redis_pipeline.hset(k, mapping=inner_dict)

        result = redis_pipeline.execute(raise_on_error=True)
        print(result)

Basically it:

1. takes in input a list of several thousands dictionaries
2. for each of those dictionary, pushes in Redis each key/value item (values are also dictionaries, that's why I'm using hset)

Each dictionary contains ~10k elements, so redis_pipeline.execute(raise_on_error=True) happens once every ~10k hset.

I noticed that after few minutes, result value step from arrays of 0s to arrays of 4s and this worries me.

On one side I expect that any error should be raised as Exception (raise_on_error=True), but on the other side I'm failing to find any reference about this behaviour in the documentation and I don't understand what does that mean.

So my questions are:

1. Does result equals to an array of 4s mean that something went wrong in the redis_pipeline.execute(raise_on_error=True) operation?
2. If yes, how can I understand what went wrong?
3. If nope, what does that mean instead?

Thanks in advance.

Answer 1

So when using the HSET command the return value is the number of fields that were added

# check if key exists
127.0.0.1:6379> EXISTS key1
(integer) 0
# add a hash with 4 k/v pairs
127.0.0.1:6379> HSET key1 a 1 b 1 c 1 d 1
(integer) 4
# Set same fields for an existing hash
127.0.0.1:6379> HSET key1 a 1 b 1 c 1 d 1
(integer) 0
# Add an additional k/v pair
127.0.0.1:6379> HSET key1 a 1 b 1 c 1 d 1 e 1
(integer) 1
127.0.0.1:6379> HSET key1 f 1
(integer) 1

so it's possible that those entries with 0 already existed in the cache and no new fields were added.