TL;DR
- My responses to PRs and issues were sluggish, especially during the pandemic. I am sorry, and thank you for sticking with us.
- We aim for reasonable, not instant, turnaround.
- Expect a quick maintainer look at every PR within 3 business days.
- Full review typically follows once CI is green and AI review items are addressed or explained.
- We will not mass-close old issues. We are revisiting them with AI assist and closing for the right reasons.
- We are formalizing labels and dashboards to make navigation easier. Details will follow in a short, separate post.
- Suggestions are welcome.
I want to be very honest with you: my responses to pull requests and issues have been sluggish for quite some time. This affected the whole rsyslog project, because in the end it always comes back to the limited capacity of the maintainers – and most often, that means me.
We are working hard to change this. It will not mean ultra-fast turnaround times, but it will mean reasonable turnaround times. We have already made important steps, and AI will play a key role in improving this going forward. Still, this is work in progress, and I welcome your suggestions.
The pandemic years
When COVID hit, everything got harder. We kept shipping releases with important fixes and enhancements, but our handling of issues suffered badly.
On a personal note: besides rsyslog I serve as a councilman in my village. During the pandemic I had to step up for my fellow citizens. This was a real-life crisis, sometimes life-challenging in physical or psychological ways. I had to make tough priority decisions. I hope I found a balance that kept rsyslog in good shape while also addressing urgent local needs. Special thanks to the Adiscon team for their support in these times.
Other team members were also not fully available due to Corona effects. The aftermath left a backlog everywhere, including our tracker. In parallel we started to explore AI. That exploration consumed time, but it was the right call and led to our AI First strategy that we are now implementing.
A heartfelt thank you to David Lang, long-term contributor and part of our core team in spirit. He kept the flame burning when things looked darkest.
Before the pandemic
We were not perfect before COVID. Response times were already slower than we wanted, but we managed them much better. Some issues from that period are still open.
A frequent pattern: requests to interface rsyslog with other systems. They looked simple to the requestors, but for us they required learning the other system well enough to integrate correctly. That learning curve was the bottleneck then, and sometimes still is. The good news: with AI First this hurdle has lessened. We are actively revisiting older integration requests, and if they still make sense in today’s environment, we will try to implement them.
Why we will not mass-close old issues
Old issues are not clutter. They capture real needs and valuable insight. That is why I object to closing everything older than a cutoff date. Instead, with AI assistance, we are reviewing them carefully and closing them for the right reasons – often by actually implementing what is needed.
How we work now
- Quick look: A maintainer usually does an initial look at each PR within 3 business days. The goal is to avoid early PRs getting lost.
- AI review: We run automated AI reviews. Instantly after the PR is opened. In our experience they are 90%+ correct and provide actionable items.
- Contributor expectations: Please follow the AI reasoning. Either fix what it flags or explain clearly why it does not apply. When AI/CI are green and something is still unclear, leave a brief comment and we will jump in; I get notifications and almost always check the current PR state very quickly (if I overlook one, a ping helps) – you can verify our pace in the repo’s recent PR activity. I usually do not comment when there are obvious issues – but I do comment if I notice contributor overlooks that.
- Full review cadence: A deeper maintainer review generally happens once CI is green and AI-raised items are addressed or reasoned about. If I see you are stuck, I will help. If I think AI feedback is being ignored, I will nudge you to re-check it first.
- Reasonable timelines: We will not promise instant reviews. We commit to reasonable turnaround and better communication when something is blocked.
AI will also help us improve response times for issues, triage, and integration research. We apply AI in a responsible AI First style: any AI tools we use for issue / PR handling must make sense and provide real value.
Policy and process work
We are formalizing labels and lightweight dashboards to make it easier to find entry points and track progress, including good-first-issue style labels. This would add too much detail for this post, so I will publish a short follow-up with specifics. Suggestions are very welcome.
What to expect next
We face an enormous mass of work and limited human resources. We will not overpromise. We will continuously refine our path forward: better triage, clearer expectations, more automation, and steady work on the backlog. If you see a place where a small improvement would make a big difference, please tell us.
I am sorry for the times when you felt ignored. I am deeply grateful that so many of you stayed with us. Rsyslog is strong, it keeps evolving, and with your help and modern tools we will make contributing smoother and faster.
Thank you for your patience, your trust, and your contributions.
Rainer Gerhards
rsyslog maintainer