[DRBD-user] DRBD 8.4.2 Doesn't Compile Against 3.7 Kernel

Lars Ellenberg lars.ellenberg at linbit.com
Thu Dec 6 13:36:03 CET 2012

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


On Tue, Dec 04, 2012 at 08:28:32AM -0600, Shaun Thomas wrote:
> On 12/04/2012 02:47 AM, Lars Ellenberg wrote:
> 
> >The next merge window (so, 3.8) should see the equivalent of 8.4.2,
> >plus patches (whatever comes up during the merge window).
> 
> Well, what do you know:
> 
> #define REL_VERSION "8.3.13"
> 
> That would explain a couple things, anyway.
> 
> >Maybe you should get your DRBD (and Support) from LINBIT instead.
> 
> Well, DRBD does seem to be a "product of LINBIT". And I was
> definitely wrong about 8.4.2 being in mainline. Not sure why I
> thought it was. But if LINBIT is the only source of DRBD, and 8.4.x
> won't compile against, and isn't included in the latest kernels,

You are aware that you complained about
released out-of-tree module code not compiling
against latest upstream *release candidate* kernel code?

Maybe you should simply use a released kernel?
Or better yet, an enterprise distribution?

If you insist on using git checkouts,
you need to be either fit to fix things yourself,
get external help to fix it,
or live with the occasional breakage.

Besides, it *does* work.
If it breaks again, we fix it.
Obviously we cannot fix it before upstream breaks, right?
So we need to notice the breakage, then fix it,
and then publicize those fixes.

Each of these steps can be delayed...

> where else would one magically obtain it?

Why would you want to obtain it elsewhere.

> Theirs is clearly broken in that regard,

That is factually wrong.

> and so far as I know, they're the only source of it.

Yes, we at LINBIT develop DRBD.

> My question can be slightly modified to apply to 8.3.13, though. If
> the code has been modified to compile against API changes in 3.7,
> it's not the same version.

We very much try to keep the out-of-tree code in sync with the in-tree
version. Obviously, as development happens out-of-tree still, *and* the
out-of-tree code works with all kernel versions back to 2.6.5 (or so), 
the out of tree code necessarily has to be different.

> Should I be worried about DRBD in general?

Sure.
If you are involved with high availability or data integrity,
you should always worry. That's part of taking it seriously.

In part it also comes down to: do you trust us to know what we are doing.
If not, we even use DRBD?

> Or at least 8.4.2? It seems like the community is ignoring it even

Is that so?

> though the DRBD site and all documentation seem to point to it as the
> preferred version.

8.4.0 was clearly a .0.
.1 was better, but still had glitches.

I'd say (not only) the community is just conservative, and rightly so.

We still have customers that use 0.7, which really is not even close to
the 8 series in "correctness", even;
but, "never change a running system"...

There is a huge number of installations still on the 8.0 branch,
which has not received changes or bugfixes for a long time already.
Because, 8.3 received new features.
Which sometimes also is the source of "unstable-ness".

Now that 8.3 is the current stable branch, I expect most conservative
people to settle on that for the time being.  Or follow 8.4,
for new deployments, performance improvements, or new features.

Or because 8.3 at some point will stop compiling against your favorite
git checkout of the upstream kernel.

Of course, you could also just contact LINBIT,
if you need specific help with DRBD or high availability issues.


-- 
: Lars Ellenberg
: LINBIT | Your Way to High Availability
: DRBD/HA support and consulting http://www.linbit.com

DRBD® and LINBIT® are registered trademarks of LINBIT, Austria.
__
please don't Cc me, but send to list   --   I'm subscribed



More information about the drbd-user mailing list