On Fri, Jan 1, 2016 at 5:36 PM, Kent Overstreet
[off-list ref] wrote:

On Thu, Dec 31, 2015 at 04:19:03PM -0500, Denis Bychkov wrote:

quoted

On Thu, Dec 31, 2015 at 12:18 AM, Kent Overstreet
[off-list ref] wrote:

quoted

On Wed, Dec 30, 2015 at 08:25:36PM -0700, Jens Axboe wrote:

quoted

On 12/30/2015 08:15 PM, Kent Overstreet wrote:

quoted

On Wed, Dec 30, 2015 at 10:59:39AM -0700, Jens Axboe wrote:

quoted

Looking over these, most are really simple one-liners, and nothing sticks
out as being overly complicated. Kent, do you have any plans to maintain the
in-kernel bcache?

Yeah - these patches are all fine, go ahead and pull.

Great, thanks.

quoted

I may start doing maintainence again at some point (but if there's someone
willing to step up and take over and do a good job of it, I'd gladly hand things
off)

As long as we have a path into mainline for stability fixes, at least that's
better than before.

I'd really like to get the improvements from the bcache-dev branch upstream -
there's a lot of _huge_ improvements (performance and otherwise), but
backporting the non on disk format changes has turned out to be... not really
practical.

So one of the major obstacles has been that there's a ton of very worthwhile
code I'd really like to get upstream, but at this point it's pretty much going
to have to be as drivers/md/bcache2 - effectively a fork that wouldn't support
the original on disk format. And that's a high hurdle.

Hey Kent,

Why is it so important to keep the same on-disk format? We are are
talking about the
caching device, not the backing device (which does not have its own
on-disk layout, it's
just the layout of the FS it backs, correct?)
So what's so big of a deal if the caching device format changes? You
just disconnect the cache set
before the upgrade, flushing all the cached data that is not on the
backing device, disable caching for
this device (bcache can work without the caching device in
write-through mode), then upgrade the
kernel and re-create the caching device with the new format. Yes, all
you cache is invalidated, but it
will take few days or, in case of very intensive use/lots of new data,
even few hours. And those who
can't tolerate this warm-up period can stick with the old code. But,
if you say there is A LOT of performance
improvements, it definitely should be worth it.
It's not like you are going to lose your backing device data, only
invalidate the cache.
So, can you please tell me where I am wrong here and why can't we do this?

We certainly can do all that, but: since new bcache can't read the old bcache
format (I can go into why that's impractical, if people are curious) - that
means there's a pretty high cost to switching to the new format:
 - people have to manually upgrade
 - the kernel would have to carry around both the old and the new
   implementations of bcache for as long as people are using the old format -
   we can't force people to upgrade

Not that hard technically, you could just leave the existing bcache
module as-is to avoid regressions when adding the new format support,
since the existing code does not require a lot of maintenance and add
a new module that would only recognize a new superblock. Another
question is how easy it is to convince Linus/top maintainers to keep 2
modules with a lot of duplication with intention to retire the old
code eventually, but something tells me that you know ways around this
problem.

So this isn't something we want to do more than once, which means we need to
make sure the new on disk format is 100% done. And it's not quite done - the
main thing that's left for it to really be considered done is big endian support
and endian compatibility (writing the code so a little endian machine can read a
big endian layout and vice versa; due to the way bkeys work it's not possible to
just have an endian agnostic layout, we'll have to do swabbing).

But this problem is not unique to bcache at all, AFAIK, almost any FS
linux supports would not be able to work on a different endianness
than it was created for.

And this isn't a trivial amount of work - and besides finishing the on disk
format, there's a fair amount of work on tooling and related stuff to make sure
everything is ready for the switch.

And, I can't work for free, so somehow funding has to be secured. Given the
number of companies that are using bcache, and the fact that Canonical and SuSe
are both apparantly putting in at least a little bit of engineering time into
supporting bcache, you'd think it should be possible but offers have not been
forthcoming.

I don't know, IMHO bcache was hurt a lot because of a host of small
problems that nobody was able to address for quite some time. It
gained a bad reputation as a production system, unfortunately, which
means not much interest from the enterprise world, which means
Canonical & co. did not want to invest into it. Don't get me wrong, I
am not blaming you. Of all people, I might understand pretty well what
was going on, just explaining why RH or Canonical or Suse did not
fight for the privilege to financially support this project.

quoted

Speaking for myself I can help with maintenance/coding/unit test
writing/code reviewing.
I realize, you have no idea about my skills, but I do have some
experience with low level/
systems programming. I don't have a lot of DEEP knowledge about linux
kernel, but I did a lot of
driver-related programming back in the day, when memory was a scarce
resource (OS/2 in 90s :).
It was long ago, I admit, but I can learn pretty quick and, besides,
can help with some trivial stuff
like regression tests/debugging, etc

That would be useful, but I've had a fair number of offers for help before but
no one has actually committed the time so far.

Well, I get your bitterness, but there is only one way to find out, right?

-- 

Denis