[DRBD-user] frequent split brains without network interruption

Christian Still c.still at netattalk.de
Fri Jun 8 10:56:53 CEST 2018


Hi all,

the following error occurs reproducible on various drbd-mirrors we are using:

drbd r0: BAD! BarrierAck #1729154 received, expected #1729153!

And it always ends in an splitbrain, as we are using dual-primary ocfs2 on these devices.

OS is Ubuntu server 16.04 x64 with 4.4.0-87-generic Kernel and:
cat /proc/drbd
version: 8.4.5 (api:1/proto:86-101)

Network link, is directly connected fiber from server to server via 10Gbit:
product: MT26448 [ConnectX EN 10GigE, PCIe 2.0 5GT/s]
vendor: Mellanox Technologies
driver=mlx4_en driverversion=2.2-1 (Feb 2014) duplex=full firmware=2.9.1000
using different storage devices (nvme SSDs;SATA AHCI SSDs, mdadm RAID10 over HDDs ..)

We think that the error occurs especially in load situations.
Network link problems, physically, can be ruled out as a cause.


Maybe anyone has an idea on this, or even solved it before.

Thanks in advance for any answers.


Regards, CS
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20180608/b75fb0f5/attachment.htm>


More information about the drbd-user mailing list