Re: rsync to ext2 filesystem issues

[Luis Falcon <falcon%gnuhealth.org@localhost>]

Good morning, Ramiro
El Wed, 16 Sep 2026 13:02:43 +0200
Ramiro Aceves <ea1abz%gmail.com@localhost> escribió:
> El sáb, 12 sept 2026 a las 1:59, RVP (<rvp%sdf.org@localhost>) escribió:
> >
> > On Fri, 11 Sep 2026, Ramiro Aceves wrote:
> >  
> > > El vie, 11 sept 2026 a las 9:29, RVP (<rvp%sdf.org@localhost>) escribió:  
> > >>
> > >> How much longer? I check my ext{2,4} data partitions using
> > >> e2fsck on both FreeBSD and NetBSD, and have seen no appreciable
> > >> difference in speeds. 
> > >
> > > On Linux, fsck took about 1 hour.
> > > On NetBSD, using fsck from e2fsprogs, it took about 4.85 hours.
> > >  
> >
> > External USB drive, you said, right? That would explain it. For
> > some reason, NetBSD is considerably slower than Linux in this. Even
> > a simple dd(1) read test'll show this. (My ext* partitions are all
> > on an NVMe drive.)
> >
> > -RVP  
> 
> Hello RVP
> 
> After several days playing with this issue, I have no conclusions but
> the crash. I do not appreciate that Linux is  faster enough to explain
> that difference in fsck performance. Perhaps NetBSD  fsck ext2
> implementation is slower than the Linux one. I do not know.
> I am about getting crazy with this ;-) It surpases me.
> 
> I have found this on the NetBSD page, I do not know if it is valid
> nowadays or not:
> 
> https://wiki.netbsd.org/tutorials/how_to_share_an_ext2_partition_with_linux/

Thank you for sharing!

> It turns out that the first one is read very slow using dd:
> 
> dd if=/mnt/datos/testfile1 of=/dev/null bs=1m about 1.4 MB/s
> dd if=/mnt/datos/testfile1 of=/dev/null bs=1m about 49 MB/s
> dd if=/mnt/datos/testfile1 of=/dev/null bs=1m about 30 MB/s
> 

We have to be cautious about the IO stats when writing to external
conventional drives (non SSD). My experience and my impression with this
type of drives is that copying is heavily influenced by the operating
system caching, dirty pages, (in the case of linux kernel take a look at
"dirty_bytes" in
https://www.kernel.org/doc/html/latest/admin-guide/sysctl/vm.html). I
have tried different filesystems (ext2, btrfs..) and although it can be
a slight difference among them, my understanding is that those
bottlenecks, random changes of IO wait queue and throughput  depend
more in the underlying technology (hardware) and the operating system
than in the filesystem itself (although is important, too)

Best,