HP-UX HB v13.00 Ch-15 - Serviceguard

HP-UX Handbook – Rev 13.00 Page 76 (of 108)
Chapter 15 Serviceguard
October 29, 2013
and to check for errors returned by the network drivers. Discussing all possible DLPI error
conditions would exceed this document's scope. It is usually best practice to check affected
interfaces with tools like linkloop(1M) and lanadmin(1M). In the past many of those problems
were tracked down to defective hardware components. Some of those problem were caused by
Serviceguard defects, so it is also advisable to install a current Serviceguard patch to address
known problems in this area.
• syslog message indicates DLPI message corruption.
Possible messages include:
cmcld: DLPI message too small (X < Y + Z). Ignoring the message.
cmcld: DLPI message checksum incorrect upon completion. Group, X, seems to be
corrupting the message.
cmcld: DLPI message not sent because oversized (X + Y > Z).
cmcld: Length of DLPI header (X) is too small. Ignoring the message.
cmcld: DLPI message too small (X < Y + Z). Ignoring the message.
cmcld: DLPI message too big (X + Y). Ignoring the message.
cmcld: The comm_link message length is inconsistant. (X < Y + Z).
cmcld: DLPI message checksum incorrect upon receipt from X. Ignoring the
message.
Beginning with Serviceguard A.10.12 DPLI traffic is protected against corruption using
checksums. Earlier revisions could abort under such conditions resulting in a node TOC. The
affected hardware (interfaces, network switch or hub, other hardware on that bridged net, etc.)
should be checked and replaced if needed. See also HA Newsletter #15 (HP internal) for details.
• Problem: Crossover cables for a Heartbeat LAN of 2-node clusters
When either LAN card fails, or the crossover cable is disconnected, both LAN cards go down.
This is because the electrical signals necessary for the cards to determine that a valid LAN
connection exists are not present. The result is that since both nodes appear to have a bad LAN
card, Serviceguard may TOC the wrong node. On multi-speed cards, such as 10/100Base-T, the
cards should not negotiate which speed will be used when the system boots up. Otherwise, if
only one system is booted and the remote system is down, the negotiation would fail, and the
card would not be enabled at all. So when the second node eventually comes up, it's LAN would
also be down.
For the reasons listed above, it is not recommend (but supported) to use crossover cables for
Serviceguard configurations. Auto-negotiation should be disabled by setting a fixed
speed/duplex configuration.
Problems with VxVM Disk Groups
This section describes some approaches to solving problems that may
occur with VxVM disk groups in a cluster environment. For most
problems, it is helpful to use the vxdg list command to display the disk
groups currently imported on a specific node. Also, you should consult
the package control script log files for messages associated with