Hi Edward, I really appreciate that you are putting attention into this part of the codebase, which is admittedly underserved.
I also want to acknowledge that it's frustrating to send up code and get no response. I sometimes have this same experience, even as an Apache Software Foundation member, when I branch out and send code to a project in which I'm not a committer. It stings a little. This isn't an excuse, just an acknowledgment that it's hard when your needs aren't served. I will not block a 3.5.0 release on these container-executor improvements. There is simply too much demand for other 3.5.0 features like full Java 17 compatibility. Your work is very well-intentioned. There is just too much demand for the rest of 3.5.0 to slow down. I'm getting hit up by multiple downstream ASF projects regularly asking for 3.5.0. For most of these projects, container-executor is not directly relevant. This isn't intended to minimize your work (which is important), just trying to balance against overall community needs. That said, let's see what we can get done. We have some time, because we don't have an adequate response to MAPREDUCE-7527 yet. (Hint: can you help there?) I've been incredibly focused on 3.5.0 blockers, but they're mostly cleared now. Your YARN-11919 is now at the top of my review queue. I'll also aim to review the rest. Again, no commitment to inclusion in 3.5.0, but I hope it helps to hear your good work is at the front of my queue. Chris Nauroth On Thu, Feb 26, 2026 at 8:17 AM Edward Capriolo <[email protected]> wrote: > (Excuse if this is a double-send as I transmitted the original before fully > joining the ML) > > Hadoop Team. I have been trying to have this conversation with you all on > the use@ lst for 1 or 2 months now, but now I realize Im like the only > person talking there :) > > https://lists.apache.org/thread/mpd29n3ckpo3qt1jw56gbog4gcn3jrly > > I have posted to several mailing lists including the SECURITY lists. I will > make the same point(s). I assert the container executor is broken, it > doesn't work as designed. It also mishandles pointers, which(email sent to > security list). > > I would be -1 (nonbinding) on a release that doesn't address the 2-3 > tickets I have put in to fix LCE. > > Im struggling to word my frustration, repeat my sentiment from the above > thread. > > Lce isnt the "cathedral" . I would hazard a guess that from the state of it > the project could use someone that WANTs to deal with it. Hint hint .. im > the only fish on the hook so real me in folks. > > Its a chicken-eggg issue. "we dont have bandwidth to review" and "the only > way to get more bandwidth is more committers", "but we cant get new > committers, because we need to get them 10 commits and have a vote" :) >
