Accepted Solution

Way for Harvest to graph inode usage?

Is there a way to graph inode usage in volumes?

Re: Way for Harvest to graph inode usage?




Capacity info is collected from OCUM and as far as I know inodes used (files) per volume is not tracked, which means Harvest can't pull it either.  If you use Quotas then Harvest does pull 'files_used'.  If your data is in qtrees then summing them up in Graphite/Grafana, grouped by volume, would give you the volume total file count.  A query like this would do it:



groupByNode(netapp.capacity.$Group.$Cluster.svm.$SVM.vol.$Volume.qtree.*.files_used, 7, 'sum')



If your data is not in qtrees, you or you want to collect it natively, you can use a custom template to add in a counter that shows inodes used "wv_fsinfo_public_inos_used" which contains the current inode count for a volume.  To do this follow these steps:


1) Create a new config template that includes this counter:



cat << EOF > /opt/netapp-harvest/template/cdot-volume-files.conf
%poller =   (
        'volume' =>
                                counter_list     => [ qw(vserver_name vserver_uuid node_name node_uuid parent_aggr
                                                                        read_data write_data
                                                                        read_ops write_ops other_ops total_ops
                                                                        read_latency write_latency other_latency avg_latency
                                graphite_leaf    => 'svm.{vserver_name}.{node_name}.{parent_aggr}.{instance_name}',
                                plugin           => 'cdot-volume',
                                plugin_options   => {'svm' => 1, 'node' => 1, 'aggr' => 1},
                                enabled          => '1'


2) Update your poller entry in /opt/netapp-harvest.conf to merge this config template on top of the default one by using the template variable like this:



hostname       =
site           = dev
template = default,cdot-volume-files.conf



3) Restart your pollers


/opt/netapp-harvest/netapp-manager -restart



4) After 2 polling cycles (120 seconds) create a graph in Grafana for a metrics path like (replace my site/cluster/svm/volume names with your own, use wildcards, or use template variables in Grafana):



In case your next question is to show % inodes used or total (potential) inode count for a volume I checked for counters for these but could not find any.  To capture these you would be best to use the volume API (Harvest only knows how to use the counter manager API from the conf template) or a shell script of df -i and parse and send these in to Graphite directly.


Hope this helps!



Chris Madden

Solution Architect - 3rd Platform - Systems Engineering NetApp EMEA (and author of Harvest)

Blog: It all begins with data


If this post resolved your issue, please help others by selecting ACCEPT AS SOLUTION or adding a KUDO or both!