[DRBD-user] BAD! .... messages

Christoph Lechleitner christoph.lechleitner at iteg.at
Tue Nov 22 09:38:02 CET 2016

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

Am 2016-11-22 um 01:08 schrieb Jan Bakuwel:
> Hi Christoph,
> Not replying to the list ...?

Ooops, that wasn't intentional.

> Do you know if the kernel_sendmsg() bug affects the 3.16 kernel too?

No, that bug only affects 4.0-4.9 (final 4.9 will have the fix, 4.10 
gets a rewrite there anyway).

> Do
> you know if that bug related to the issue I reported?

I don't think so.

Main reason I jumped in was to mention our special Debian repository as 
easy and free access to the latest released DRBD 8 kernel module:

 >> When using 4.x kernel from jessie-backports or the like, and as long 
as a certain bug fix does not reach those kernel packages, I invite you 
to try ours, as described in https://confluence.clazzes.org/x/CAC2

> I'll set up a virtual lab with two Jessie systems running the stock 3.16
> kernel and see if your drbd8-dkms and drbd-utils package work.

That'd be great to know, Thanks.

> Upgrading the kernel as well is just a bit too uncertain for me at this
> stage...

I'm with you there, we also don't touch the kernel version unless we 
have to.

>> AFAIK $$$$.
> Yeah. I don't mind paying a fair contribution, I like it too when people
> pay for my time.  I could justify $ (maybe even $$) but certainly not
> $$$$ as I'm running a small ICT business mainly working for
> not-for-profits (and one small business). I maintain a total of 5 pairs
> of servers (2 of those pairs are Alix APU's) and am running production
> on all those.

Sounds like you are just in our league.

Paying an expert (whom we knew from the regular dinners of our local 
Linux User Group) did cost only half of what Linbit wanted per year per 
node ;-)

Business Remark: I fully understand why Linbit charges $$$$, and I hear 
the only similar product, CephFS?, charges even $$$$$$.
So no hard feelings, but we found a cheaper and even more local solution 
and share the results because we try to live OpenSource.

 > Can't really use those for experiments.

We reactivated an obsolete node pair for that, only to learn we weren't 
able to reproduce our problem there.
(Again, our problem only hit kernels 4.0-4.9 and is fixed now)

> Thanks for your reply and also for taking the effort and $$$ ($$$$?) to
> have the bug fixed, as well as making the packages available!

Fortunately we only had to spent $$, but yes our limit was somewhere at 

Regards, Christoph

More information about the drbd-user mailing list