2010-08-23 12:05 PM
I am wondering if there is a way for DFM to help me trigger an alarm when a hosts's filesystem (netapp non-thin-provisioned lun) is almost full. I only seem to be able to set these thresholds on volumes and for a number of reasons I am looking to see this managed at the lun logical object level.
Use case might be something like I have an MSCS cluster with host agents installed on both nodes. Nodes have netapp luns (space reserved) mapped. When I failover the snapinfo for one of my snapmanager jobs doesn't clear the old information from my primary node and hence the snapinfo disk fills up after a number of snapshots. I"m looking for a way to tell that I may have a disk filling which can lead to an application outage.
Under the hostagent alarms I don't see anything related to this. Only that a host agent has been discovered, path has been discovered, iscsi/fcp service is down, etc... I also don't see alarms related to luns for space available/full. the only trigger I can find at all relates to the management station or performance advisor disk maps being full or empty.
I'm assuming I need the host agent involved here since nothing else in the stack has any ownership or understanding of what the lun object contains.
2010-08-23 04:32 PM
Unfortunately, you can't use DFM to alert you when a host agent's file system fills-up. The FSRM feature doesn't specifically capture the file system "percent full" as part of its counters. There are a number of open systems host monitoring utilities that suppy that functionality (Nagios, etc...) and DFM isn't trying to reinvent that wheel.
ONTAP and DFM cannot peek into a LUN and interpret the file system, so they can't tell you how full it is. Only the host O/S that is writing to the LUN knows that fact. So that's why you don't see any events or alarms related to "LUN full".
However, you could script your own alarm here. When used with the File SRM feature, the NetApp host agent reports how many bytes of data are in the entire path (dirSizeTotal). You could divide this number by the total capacity of the file system to calculate the percent-full. If its over a certain amount, you could e-mail yourself.
To get the total bytes in a path, use the following command from the DFM server CLI:
# dfm srm dir details <FSRM PATH>
# dfm srm dir details demo-esx-opsmgr:E:\
dirOldestFileModified 17 Nov 2009
dirFileAverageAccessed 12 Aug 11:44
2010-08-23 11:13 PM
You can use the below cli to generate your custom event for the script, in DFM.
# dfm event generate help
generate -- generate a particular user-defined event
dfm event generate [ -t <timestamp> ] <event-name> <source> [ <event-message> ]
This command is used to generate custom events. Script plugins can
use the command to generate built-in script status events such as
script:warning-event. This command cannot be used to generate other
timestamp: specifies the event generation time in
If not specified, the time when the command is invoked is used.
event-name: specifies the name of the event.
source: specifies ID/name of source object of the event.
event-message: A message specific to this event.
This message will be displayed as part of event details.