What if application teams use Redis commander to manually insert/update data in Redis Sentinel Cluster.
I think in a well designed system, the application should be the thing writing to Redis and Redis commander exists mainly read only inspection tool.
We've been running Nexus Repository OSS/CE for a few years as our central proxy + hosted repo (Maven, npm, Docker, PyPI). Since the Community Edition limits tightened, we're bumping the ceiling regularly. Our CI is the main culprit: a handful of parallel builds pulling through the proxy chews through the daily budget fast, and once you're over, publishing gets blocked until the rolling 24h count drops back under.
Pro pricing isn't in the budget right now, so I'm looking at what else is out there before we commit to anything.
What I'm considering:
Harbor — solid for OCI/Docker/Helm, but doesn't cover Maven/npm/PyPI, so we'd need something alongside it.
Pulp 3 — multi-format and genuinely open source, but looks heavier to operate (Django + workers + Redis + Postgres).
JFrog Artifactory OSS — free tier is basically Maven/Gradle only.
Gitea / Forgejo package registry — attractive if you're already self-hosting the forge; not sure how it holds up as a caching proxy at scale.
Questions for anyone who's actually done this:
What did you migrate to?
How painful was moving existing hosted artifacts?
For anyone who stayed on CE — did aggressive cleanup policies and a caching layer in front (nginx/Varnish) actually get you back under the limits, or is that just delaying the inevitable?