Mailing List Archive: DRBD: Users

drbd-8.3.3rc3.tar.gz

Index | Next | Previous | View Threaded

philipp.reisner at linbit

Sep 29, 2009, 6:00 AM

Post #1 of 3 (582 views)
Permalink

drbd-8.3.3rc3.tar.gz

Hi,

We are still holding our breath how and when DRBD will land in
Linux mainline. I was told, that it can go in after 2.6.32-rc1
since it is a driver and not a patch to the more central parts
of the kernel. It will go in through Jens Axboe, although I
do not know when exactly.

To relieve the strain a bit, here is an other rc of DRBD-8.3.3

The main difference to rc2 is that the default values of
sndbuf-size rcvbuf-size where changed to 0 (auto tune).
They have been 128k before.

This take the burden to get the socket send buffer tunes
right off the administrator, but make sure that TCP's sysctls
settings are sane. I.e.

net.core.rmem_max = 10485760 (byte per socket)
net.core.wmem_max = 10485760 (byte per socket)
net.ipv4.tcp_rmem = 65536 87380 10485760 (min/def/max byte per socket)
net.ipv4.tcp_wmem = 65536 65536 10485760 (min/def/max byte per socket)
net.ipv4.tcp_mem = 128000 170666 256000 (min/pressure/max pages in total)

8.3.3rc3 (api:86/proto:86-91)
--------
* Correctly deal with large bitmaps (Bugz 239, 240)
* Fixed a segfault in drbdadm's parser for unknown sync-after dependencies
* DRBD_PEER was not set for handlers (introduced in 8.3.2) (Bugz 241)
* Fixed a bug that could cause reads off diskless DRBD devices to get very slow
* Fixed a deadlock possible when IO errors occure during resync (Bugz 224)
* Do not do a full sync in case P_SYNC_UUID packet gets lost (Bugz 244)
* Do not forget a resync in case the last ACK of a resync gets lost
* The UUID compare function now handles more cases when connection/disk got
lost during UUID updates (Bugz 251, 254)
* If a resource gets renamed (only) update its /dev/
* drbdsetup get-gi/show-gi sometimes warned about unknown tags (Bugz 253)
* Autotune sndbuf-size and rcvbuf-size by default
* Fixed many spelling errors
* Improvements on the crm-fence-peer Pacemaker integration
* Do not upgrade a Consistent disk to UpToDate when the fence-peer handler
can not reach the peer (Bugz 198)
* Support for Infiniband via SDP (sockets direct protocol)
* Install bash completion stuff on SLES11
* Following Linux upstream changes 2.6.31

http://git.drbd.org/?p=drbd-8.3.git;a=tag;h=drbd-8.3.3rc3
http://oss.linbit.com/drbd/8.3/drbd-8.3.3rc3.tar.gz

--
: Dipl-Ing Philipp Reisner
: LINBIT | Your Way to High Availability
: Tel: +43-1-8178292-50, Fax: +43-1-8178292-82
: http://www.linbit.com

DRBD(R) and LINBIT(R) are registered trademarks of LINBIT, Austria.

_______________________________________________
drbd-user mailing list
drbd-user [at] lists
http://lists.linbit.com/mailman/listinfo/drbd-user

gernot.schmied at chello

Oct 2, 2009, 2:45 AM

Post #2 of 3 (513 views)
Permalink

Re: drbd-8.3.3rc3.tar.gz [In reply to]

Am Dienstag, den 29.09.2009, 15:00 +0200 schrieb Philipp Reisner:
> Hi,
>
> We are still holding our breath how and when DRBD will land in
> Linux mainline. I was told, that it can go in after 2.6.32-rc1
> since it is a driver and not a patch to the more central parts
> of the kernel. It will go in through Jens Axboe, although I
> do not know when exactly.
>

Hi Linbit folks!

for what it's worth:

I was following the discussion on the Linux kernel mailing list and the
strange arguments about unified frameworks with RAID and LVM and the
user space control tool discussion. Congratulations for the patient and
calm way of replying and hanging in in the face of very awkward
arguments brought up against DRBD inclusion.
I got the impression that the majority of kernel maintainers realize the
"pressure" of an established solution that is used in production all
over the world and the huge leverage of distribution builders who
require DRBD to fill a gap. At the end of the day deployment base and
user acceptance for an elegant and reliable solution will surely make
the difference.

Gernot

_______________________________________________
drbd-user mailing list
drbd-user [at] lists
http://lists.linbit.com/mailman/listinfo/drbd-user

lars.ellenberg at linbit

Oct 2, 2009, 3:37 AM

Post #3 of 3 (497 views)
Permalink

Re: drbd-8.3.3rc3.tar.gz [In reply to]

On Fri, Oct 02, 2009 at 11:45:14AM +0200, Gernot Schmied wrote:
> Am Dienstag, den 29.09.2009, 15:00 +0200 schrieb Philipp Reisner:
> > Hi,
> >
> > We are still holding our breath how and when DRBD will land in
> > Linux mainline. I was told, that it can go in after 2.6.32-rc1
> > since it is a driver and not a patch to the more central parts
> > of the kernel. It will go in through Jens Axboe, although I
> > do not know when exactly.
> >
>
> Hi Linbit folks!
>
> for what it's worth:
>
> I was following the discussion on the Linux kernel mailing list and the
> strange arguments about unified frameworks with RAID and LVM and the
> user space control tool discussion. Congratulations for the patient and
> calm way of replying and hanging in in the face of very awkward
> arguments brought up against DRBD inclusion.
> I got the impression that the majority of kernel maintainers realize the
> "pressure" of an established solution that is used in production all
> over the world and the huge leverage of distribution builders who
> require DRBD to fill a gap. At the end of the day deployment base and
> user acceptance for an elegant and reliable solution will surely make
> the difference.

Thanks.

Linus: "I'm perfectly happy to pull drbd,
but please do it for the next merge window."
http://lkml.org/lkml/2009/10/1/385

--
: 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
_______________________________________________
drbd-user mailing list
drbd-user [at] lists
http://lists.linbit.com/mailman/listinfo/drbd-user

Index | Next | Previous | View Threaded

Interested in having your list archived? Contact Gossamer Threads