[drbd-mc] 7 Issues Identified in drbd-mc-0.4.1

Mike Peachey mike.peachey at jennic.com
Wed Nov 11 16:55:53 CET 2009


Hi, People.

The following issues have been encountered while using drbd-mc from the
pre-compiled 0.4.1 jar file.

The jar file has been run using:
$ java -jar /path/to/DMC-0.4.1.jar --keep-helper

on Slackware-13.0 using:

java version "1.6.0_16"
Java(TM) SE Runtime Environment (build 1.6.0_16-b01)
Java HotSpot(TM) Server VM (build 14.2-b01, mixed mode)

The "--keep-helper" has been specified after being advised by rasto to
modify the drbd-gui-helper script with a tweak to prevent it complaining
about being unable to find /dev/sdb which is a backing resource for drbd
on each node.

The DRBD nodes are identical and running Slackware64-13.0 with
drbd-8.3.6 and Pacemaker-1.0.6 all compiled from source using the
SlackBuilds that I am now maintaining for these applications.

In each case passwordless RSA authentication is being used to connect to
the nodes.


########################################################################
###########
# ISSUE 1 #
###########
Immediately after running drbd-mc, before loading the saved cluster
configuration, there is a tree view on the left hand side of "All Nodes"
which lists both of my available nodes. If I click on either node, I am
presented with a NullPointerException:

APPERROR: AppError.Text
release: 0.4.1

uncaught exception
java.lang.NullPointerException
null
drbd.gui.HostBrowser$HostInfo.getGraphicalView(HostBrowser.java:863)
drbd.gui.Browser.getInfoPanel(Browser.java:191)
drbd.gui.ViewPanel$4.run(ViewPanel.java:204)
java.awt.event.InvocationEvent.dispatch(Unknown Source)
java.awt.EventQueue.dispatchEvent(Unknown Source)
java.awt.EventDispatchThread.pumpOneEventForFilters(Unknown Source)
java.awt.EventDispatchThread.pumpEventsForFilter(Unknown Source)
java.awt.EventDispatchThread.pumpEventsForHierarchy(Unknown Source)
java.awt.EventDispatchThread.pumpEvents(Unknown Source)
java.awt.EventDispatchThread.pumpEvents(Unknown Source)
java.awt.EventDispatchThread.run(Unknown Source)

###########
# ISSUE 2 #
###########
During the load sequence when choosing to load the saved cluster
configuration, the last message that flashes up is the red text:
"Heartbeat Status Failed".


###########
# ISSUE 3 #
###########
Immediately upon loading the configuration, and every 20 seconds or so
there-after throughout the life of the application the console reports
these errors:
APPWARNING: st-node1: could not find hb script
APPWARNING: st-node2: could not find hb script

st-node1 and st-node2 are stonith resources specified using the
external:stonith/ipmi resource agent (in each case using lanplus).

###########
# ISSUE 4 #
###########
In the same way as it was complaining about being unable to select
/dev/sdb before the modification to drbd-gui-helper on each node, the
console reports these errors at load time:

APPWARNING: cannot select: /mnt/store/nfs
APPWARNING: cannot select: /var/lib/nfs
APPWARNING: cannot select: none
APPWARNING: cannot select: /mnt/store/files

/mnt/store/nfs is a mount-point for an LVM2 logical volume.
/mnt/store/files is a mount-point for an LVM2 logical volume.
/var/lib/nfs is a mount-point for a bind mount of /mnt/store/nfs
I do not know to what "none" will be referring.


###########
# ISSUE 5 #
###########
In "Cluster" -> "Services" -> "Master/Slave Set (ms-store_drbd)",
drbd-mc believes that my master-slave resource (ms-store_drbd - backed
by a resource called store_drbd on each node) is not running on either
node. The resource is in fact perfectly healthy and master on one node,
while slave on the other.

The grey box for the resource has the word "starting..." in the top left
hand corner.

In the "Storage (DRBD)" view from the options tree, however, it
correctly reports that /dev/drbd0 exists on each node, that both are
Connected and UpToDate, and that each are in either the Primary or
Secondary state.


###########
# ISSUE 6 #
###########
In the "Storage (DRBD)" view, drbd-mc correctly reports the non-drbd
partitions of "/boot (/dev/sda1)" and "swap (/dev/sda3)" as mounted, but
 reports "/dev/sda2" as not mounted. "/dev/sda2" is in fact the "/"
partition for the Operating System.

In each case "/dev/sda" is shown in its own box and marked as
not-mounted. I assume this is expected behaviour.


###########
# ISSUE 7 #
###########
Related to Issue 6; the node that is the Primary DRBD node has access to
an active LVM layout on the DRBD block device "/dev/drbd0". The Logical
Volumes in the Volume Group "vg0" are "/dev/vg0/nfs" and
"/dev/vg0/files" and both are shown in the "Storage (DRBD)" view under
the master node, however both are also shown as not mounted.

Each is currently mounted on the relevant "/mnt/store/{files,nfs}"
mount-point.

########################################################################

I think that's enough for now...
-- 
Kind Regards,

__________________________________________________

Mike Peachey, IT
Tel: +44 114 281 2655
Fax: +44 114 281 2951
Jennic Ltd, Furnival Street, Sheffield, S1 4QT, UK
Comp Reg No: 3191371 - Registered In England
http://www.jennic.com
__________________________________________________


More information about the drbd-mc mailing list