On Wed, Feb 13 2019 at  4:14am -0500,
Vojtech Juranek <vjuranek@redhat.com> wrote:

> Hi Mike,
>
> >
> > Nir Soffer <nsoffer@redhat.com> wrote:
> > >    We working on enabling 4k block size in oVirt block storage domain,
> > >    built
> > >    using VG
> > >    on multipath devices on shared storage.
> > >   
> > >    We have incomplete support for 4k, added in 2011, for this bug:
> > >        [1]https://bugzilla.redhat.com/732980
> > >   
> > >    When creating or extending a VG, we check that all PVs are using same
> > >    logical and
> > >    phyisical block size, and we store both logical and physical block size
> > >    in
> > >    the VG tags.
> > >    We get the block sizes from
> > >    /sys/block/dm-X/queue/{logical,physical}_block_size.
> > >    We also enforce that device physical block size is not smaller than
> > >    logical block size,
> > >    This check was added in this patch, trying to enable block size != 512.
> > >    There is no
> > >    explanation in the patch or in the review comments why we need to
> > >    validate
> > >    this.
> > >   
> > >    [2]https://github.com/oVirt/vdsm/commit/7e79153705891a91a06eb31cd642fb2
> > >    09d10ff86 When we start to use a VG, we validate that all the devices
> > >    are using the stored logical
> > >    and physical block size.
> > >    In vdsm itself, we use the logical block size to manage vdsm metadata,
> > >    assuming that writing
> > >    and reading one block of logical block size bytes is atomic, and we can
> > >    read and write
> > >    different blocks from different hosts at the same time.
> > >    The relevant code validating PV block sizes is here:
> > >   
> > >    [3]https://github.com/oVirt/vdsm/blob/8b043e402f41d8a82b9f832be5f582b85
> > >    20b38bc/lib/vdsm/storage/lvm.py#L1110 Reading the comments in bug
> > >    732980, I don't see anything about physical block size. It looks
> > >    like this is unnecessary check, and we should check only the logical
> > >    block
> > >    size.
> > >    Regarding mixing devices with different logical block size, according
> > >    to
> > >   
> > >        [4]https://bugzilla.redhat.com/show_bug.cgi?id=732980#c8
> > >   
> > >    We should not extend an LV over devices with different block size, as
> > >    this
> > >    will change the device
> > >    logical block size (e.g change from 512 to 4k), and the change may
> > >    break
> > >    the upper layer that
> > >    already use the device and assume the previous logical block size.
> >
> > This idea that 4K writes to a 512b physical drive aren't going to be
> > atomic, and that that is going to be the basis for some upper level
> > failure is handwaving and overly paranoid TBH.
> >
> > >    Based on this, I think we are ok with limiting VG to devices with same
> > >    logical block size, so any
> > >    LV can be extended to any device.
> > >    I think this code should change to:
> > >    1. When creating a VG, check that all PVs use the same logical block
> > >    size
> > >    2. Store the logical block size in the VG tag
> > >    3. When extending the VG, check that the new PVs use the same logical
> > >    block size
> > >    4. When starting to use a VG, check that stored logical block size
> > >    matches
> > >    PVs logical block size
> > >    What do you think?
> >
> > I think you shouldn't care.  Or please show me a case where all this
> > concern matters.
>
> I'm sorry, but I'm still quite confused what needs to be checked and what not.
>
> In [1] you wrote
>
> "So the appropriate VDSM constraint is to not allow a larger
> logical_block_size device (4K) to be added to a VG that has only ever
> contained small logical_block_size (512b) devices."
>
> and
>
> "If an LV is already in use then the admin needs to avoid extending the LV in
> a way that upper layers may get upset with."
>
> and here that we shouldn't care. Could you be please more specific what one
> needs to check (regarding block sizes) when creating or extending VG and start
> using it?
>
> Thanks
> Vojta
>
> [1] https://bugzilla.redhat.com/show_bug.cgi?id=732980

Ha, only going back 8 years in the archive for that BZ!

I'd need to revisit all the details of what VDSM/oVirt are so concerned
about relative to just _always_ using 4K for the sanlock volumes.

My contention is the constraint likely wasn't ever _really_ needed.  But
maybe it was.. again, I'll look back at the BZ in more detail to see
what I'm missing.

Concerns about 4K issued to 512b physical devices _not_ being atomic
(could have 5 of the 8 512b written, so old 3 bytes could cause
issues).  IIRC I shared those concerns with Martin Petersen before
(Martin is an upstream Linux SCSI maintainer) and he felt the atomicity
concerns were overstated.  Thinking now, it was possibly for devices
that advertise 4K physical and 512b logical.  Whereas issuing 4K to a
512b/512b device could easily not be atomic for that 4K IO.

I can revisit this with Martin.  Also, I'm happy to adjust my
understanding based on further anecdotal real-world evidence that
issuing 4K IOs to a 512b device and expecting any 4K IO operation to be
atomic is _wrong_.