On Sat, 29 Aug 2020 at 10:07, Ben Kochie <sup...@gmail.com> wrote: > I think that PR link is incorrect. > > Yes, I think we should consider dropping 32-bit official builds. Unless we > can reduce the amount of mmap'd data. >
The amount of mmapping we do is only going to increase. > > One thing I noticed when running on a 32-bit machine a while back was that > the mmap'd data is mapped multiple times for different OS threads. Setting > GOMAXPROCS=1 > helped, but only a little. > That doesn't sound right, we do map some files more than once (particularly during compaction) but not per thread. > There was some discussion about doing mmap on demand, but there was not > enough interest in 32-bit to consider spending the time. > Even if all that complexity was added and the lack of predictable performance was acceptable, you'd still be limited to rather small data sizes. I don't think there's much point complicating the codebase for the sake of a Prometheus that can't handle more than a few hundred MB of data. Brian > > On Fri, Aug 28, 2020 at 3:05 PM Julien Pivotto <roidelapl...@prometheus.io> > wrote: > >> Dear developers, >> >> Based on this comment from Brian, should we stop providing official 32 >> bit builds >> of Prometheus? >> >> > 2.19 requires more virtual memory than previous versions, you need a >> > 64bit machine. 32bit system are generally not supported as we use mmap >> > a lot, if you want to try to get it to work you can but we don't >> > promise anything there as trying to support 32bit systems would >> > cripple things for the vast majority of our users on 64bit systems. I >> > can only suggest looking at Prometheus 1.x, which did not use mmap. >> >> https://github.com/prometheus/prometheus/pull/7686#issuecomment-682534007 >> >> If we build them but not recommend them, that seems like a waste of time >> and resources. >> >> -- >> Julien Pivotto >> @roidelapluie >> >> -- >> You received this message because you are subscribed to the Google Groups >> "Prometheus Developers" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to prometheus-developers+unsubscr...@googlegroups.com. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/prometheus-developers/20200828130516.GA1428921%40oxygen >> . >> > -- > You received this message because you are subscribed to the Google Groups > "Prometheus Developers" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to prometheus-developers+unsubscr...@googlegroups.com. > To view this discussion on the web visit > https://groups.google.com/d/msgid/prometheus-developers/CABbyFmrkuKc-VHAxydAHEO3PwA95V_tBu0WuuUQ-7c654mS%2Bzg%40mail.gmail.com > <https://groups.google.com/d/msgid/prometheus-developers/CABbyFmrkuKc-VHAxydAHEO3PwA95V_tBu0WuuUQ-7c654mS%2Bzg%40mail.gmail.com?utm_medium=email&utm_source=footer> > . > -- Brian Brazil www.robustperception.io -- You received this message because you are subscribed to the Google Groups "Prometheus Developers" group. To unsubscribe from this group and stop receiving emails from it, send an email to prometheus-developers+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-developers/CAHJKeLqHT724Uj9%3D7KiWZg%3DWuRwEaaOFZEZxb6TPF-gD9JkuMw%40mail.gmail.com.
