Lack of configurability for the kafka publishers, which are particularly useful for higher payloads
Adding a backlog option for the internal zmq configurations, since the default is somewhat low
the python3 kafka producer only accepts bytes and the zmq only accepts strings. So we had to encode the strings in the kafka publisher
The "metrics_include_attributes" were new are really heavy, so they were set to False by default (people can still turn them on) - prometheus might not like this new inflow so we should just turn them on if needed.
changed some logs from info to debug because they were not adding a lot of value when in info and ended up just spamming the logs
We have tested these changes on our side (although the kafka opts were hardcoded in a patch and not added thgough the configs)
The docs here were updated to contemplate these changes too.
Related to this issue: https://github.com/napalm-automation/napalm-logs/issues/382
Summing up, this PR fixes several problems:
We have tested these changes on our side (although the kafka opts were hardcoded in a patch and not added thgough the configs)
The docs here were updated to contemplate these changes too.
Let me know what you think @mirceaulinic 😄