[DRBD-announce] drbd-9.0.18-1

Philipp Reisner philipp.reisner at linbit.com
Fri May 24 13:57:13 CEST 2019


As usual this release brings a number of bug fixes, that alone is
reason enough to recommend the upgrade to every user.

Exciting are the two new features:

Quorum tie breaker
  So far enabling quorum only made sense if you have a three way
  redundant data set. With the new quorum tie breaker feature
  it works with 2 way redundnt data, but needs a 3rd, diskless node
  in the DRBD resource.

optimized meta-data access if meta-data is on PMEM
  PMEM (or NVDIMMs) is an exciting revolutionary change in computing.
  It is a new storage tier that is connected as memory to the system.
  One consequence of that is that software not gets the ability to
  access/change/update data on the storage is smaller units than whole
  sectors (512 bytes). We can not modify it in units of 64 bytes.
  That gives us a DRBD developers new possibilities in optimizing
  updates of our on-storage meta-data data structures.
  We created a new PMEM optimized layout for the activity log, and
  access the bitmap directly in place, as PMEM is memory mapped. As
  a side consequence this reduces DRBD's memory consumption.
  Depending on the workload you might get a 2x improvement in 
  write IOPs and a 4x improvement in throughput.

9.0.18-1 (api:genl2/proto:86-115/transport:14)
 * Fix an IO deadlock under memory pressure
 * Fix disconnect timing in case the network connection suddenly
   drops all packets
 * Fix some misbehavior that surfaced with Ahead/Behind
 * Fix potential spinlock deadlock in IRQ
 * Minor fixes: forget-peer, _rcu iterators
 * Quickly stop resync during AHEAD/BEHIND by introducing new
   packet for that purpose.
 * The quorum feature can now use the connectivity to
   the majority of Diskless nodes as tiebreaker
 * Access meta-data using DAX if it is on persistent memory
   (NVDIMM or PMEM); For write intense workloads this is a x2 to x4 speedup!


best regards,

More information about the drbd-announce mailing list