HP-UX iSCSI Software Initiator Support Guide (Edition 7)

If you notice one instance of iswd, it means that currently iswd daemon is active but the I-T nexus
is not active with any of the targets. If multiple instances of iswd are present, that implies that there
is at least one I-T nexus that is currently active.
If the iswd daemon is accidentally terminated the behavior displayed depends on the state of the
iswd daemon. The most likely scenario would be:
1. An attempt to terminate (kill) the parent instance of iswd will have no effect initially, because
that process is sleeping in the kernel waiting for connection open requests. The active
outstanding connections will not be affected as long as that process is in kernel space.
2. The signal sent by the kill(1) command is queued and serviced once the parent process
goes to user space upon an enqueue of a connection open request, causing all instances of
iswd to be terminated. A Connection open request is enqueued when a command requiring
iSCSI target access is executed.
The less likely scenario would be
1. An attempt to terminate (kill) the parent iswd process when it is in user space will cause all
the instances of iswd to be terminated immediately.
ACTION: Restart the iswd daemon in both the above scenerios
An attempt to terminate (kill) the child iswd process results in immediate termination of all TCP
connections to the target. If no attempt was made earlier to terminate (kill) the parent process, the
child process automatically re-spawns and TCP connections are re-established. No action is needed
in this case.
If you need to terminate iswd daemon (for example, to implement a workaround as mentioned in
the release notes), first issue “kill” to the parent process, followed by issuing “kill” to the child
process. Note that this approach requires at least one active I-T nexus.
The following message will be displayed during execution of shutdown(1m) or reboot(1m).
reboot: CAUTION: some process(es) wouldn’t die
This message is the result of the iswd daemon being kept alive to complete various tasks during
reboot. When iSCSI is configured in a system, this message may be ignored.
6.4 Diagnostics
By default, the iSCSI Software Initiator logs all diagnostic messages to the Support Tools Manager
(STM) log files. At the time of this publication, STM does not support displaying iSCSI driver logged
messages. The messages will display in STM when iSCSI support is added in a future release.
When iSCSI support is available in STM, you can configure EMS to automatically notify the system
administrator when diagnostic messages are logged by the iSCSI Software Initiator.
In addition to logging all diagnostic messages to the STM log files, the iSCSI Software Initiator
logs some of the diagnostic messages to the /var/adm/syslog/syslog.log file. For more details
on diagnostic messages, see Appendix C (page 47).
If a problem cannot be resolved using the troubleshooting techniques listed in this chapter, provide
the log file generated by the iscsidiag tool to HP Support.
iscsidiag is an iSCSI Software Initiator debug information gathering tool.
Execute /opt/iscsi/bin/iscsidiag to capture the debug information.
The logfile will be placed in the /tmp/iscsidiag directory with a filename of:
iscsidiag.<pid>.log
Provide the log file iscsidiag.<pid>.log to HP Support for further assistance. The log file
will be located at /tmp/iscsidiag.
6.4 Diagnostics 35