[DRBD-announce] drbd-9.0.0rc1.tar.gz & drbd-utils-8.9.2rc2.tar.gz

[DRBD-announce] drbd-9.0.0rc1.tar.gz & drbd-utils-8.9.2rc2.tar.gz

[Philipp Reisner]

Hi,

In the last months the focus on the kernel side was to get
it into shape to form a usable base for drbdmanage and
the OpenStack cinder driver for drbd.

It is time to share the results of this work with a wider
audience. 

We are testing here mainly on CentOS 6 VMs. Mirroring
between up to 8 nodes works usually. Usually no kernel crashes,
no refcount errors (module unload works), no protocol errors,
resyncs run in the right direction and do the right blocks.

Known issues:
 * It breaks a lot on pre-empt kernels. (Usual distros not affected)
 * Concurrent drbdadm down seems to trigger issued
 * Still possible to deadlock in the write path on a SyncTarget/Primary
   node. Though does not happen frequently.
 
The fitting drbdmanage will be announced in a few hours
today.

Please note that drbd-utils-8.9.2rc2 has to be used with the
drbd-9.0.0rc1 kernel driver. Meta data format for 9.0 was changed
again. I.e. meta-data created by older tools are not compatible
with the 9.0.0rc1 kernel driver.


drbd
9.0.0rc1 (api:genl1/proto:86-110)
--------
 * some work on drbd-9.0 to drbd-8.4 compatibility
 * many bug fixes to all areas of the new code; especially the
   the resync procedure, which is a lot more complex than it's
   8.4 ancestor
 * a transport abstraction layer that will allow for alternative
   transports besides TCP. Planed are RDMA/Infiniband and SCTP

http://oss.linbit.com/drbd/9.0/drbd-9.0.0rc1.tar.gz
http://git.drbd.org/gitweb.cgi?p=drbd-9.0.git;a=tag;h=refs/tags/drbd-9.0.0rc1


drbd-utils 
8.9.2rc2
--------
  * drbdadm 9: Log errors to syslog if stderr is not available
  * init script: fixes for stacked resources
  * fix regression corner cases in bitmap size calculation
  * allow create-md to initialize peer-max-bio-size to 1M
  * make transport selectable
  * fix aggregating drbdsetup / drbdmeta exit statii
  * some documentation fixes (content and build)
  * added direct-connect command
  * incompatible drbd-9 metadata format changeuse: node_id as index for
    peer_md instead of the bitmap_index
  * drbdadm/meta/usage_cnt: ensure output is visible
  * drbdsetup: fix arguments for all commands expecting a peer_device
  * exit codes: redefine E_USAGE to 1 (not 3)
  * some build changes
  * init script: on start, first try to load the module
  * drbdsetup events2: Improve how timestamps are assigned

8.9.2rc1
--------
  * udev rules (symlinks in /dev/by-res and by-disk) got fixed
  * Fixed upper limit for drbd-8.4 activity log entries
  * many fixes to drbdadm adjust and proxy commands for drbd-9
  * rhcs_fence: Do not invoke fence agents in parallel, rewrite in bash
  * drbdsetup events2 is now also available in drbd-8.4 (backported from drbd-9)
  * reorganized the repository have common code for drbd-9, drbd-8.4
    and drbd-8.3 only once
  * Fix drbd.ocf for resources without volume 0

http://oss.linbit.com/drbd/drbd-utils-8.9.2rc2.tar.gz
http://git.drbd.org/gitweb.cgi?p=drbd-utils.git;a=tag;h=refs/tags/v8.9.2rc2

Best regards,
 phil

[DRBD-announce] drbdmanage v0.20 pre-release

[Robert Altnoeder]

Hi everyone,

along with the rc1 release of DRBD 9.0.0, we are releasing
another pre-release of drbdmanage.

Here are some of the new features of this release:

- drbdmanage volumes are created with specified net size
   now; drbdmanage calculates the space required for
   DRBD meta-data and adds it to the space required for
   the actual data

- this also works the other way around: drbdmanage
   can report the net amount of free space, so that when
   you create a new DRBD volume, you will have that much
   space for actual data - because the space required for
   the resource's meta-data had already been considered
   when drbdmanage reported the amount of remaining
   free space

- there is a "free-space n" command now; it tells you
   the maximum size for an n-tuple redundant volume

- volumes created by drbdmanage do no longer stay
   in the "Primary" role on one of the drbdmanage nodes
   after initialization; you can begin to use them immediately

- creating a new drbdmanage cluster has even more
   automation built in now;
   if ssh keys are present, drbdmanage can attempt to
   run the join command on the newly added node
   remotely through ssh, so you are no longer
   required to copy&paste the join command to the
   new node

- more comfortable usage: most list output is
   sorted now, the size argument of the "new-volume"
   command takes unit suffixes (e.g., "35gb"), and there
   are shortcut aliases for many commands
   (for example, try "nn" instead of "new-node")

If you have questions, or if you would like to provide feedback,
feel free to post to the mailing list, or to visit our IRC channel
at FreeNode #drbd9+drbdmanage.

Best regards,
Robert