Anyone interested in NetApp 8.2 templates - CDOT and 7-mode?

[vgrig]

Working from old files from this forum (cacti_graph_template_netapp_qtree_bytes_v2.xml, netapp-qtrees-v2.xml, etc.) i managed to make QTree usage graphs work (64-bit and 4-digit volume index for CDOT).
It's an ugly hack though, so - if anyone is interested in polishing the files and templates, please let me know and i'll post them.

Some graphs are attached (one for 7-mode one for CDOT)

[rianna]

Please, post them.

[vgrig]

Attached are following files inside cacti_templates_netapp-64bit-cdot.tar.gz:
- netapp_interfaces.xml (drop it in cacti/resource/snmp_queries, you may already have this one there is you setup netapp in cacti before)
- netapp-volumes.xml (drop it in cacti/resource/snmp_queries, you may already have this one there is you setup netapp in cacti before)
- netapp-qtrees-v2-64-bit-cdot.xml (drop it in cacti/resource/snmp_queries, this is the modified netapp-qtrees-v2.xml with OID/REGEXP changed for 4-digit volumes in CDOT and some OIDs changed to 64-bit ones)
- netapp-qtrees-v2.xml (original qtree snmp quiry fils - just for reference, diiff it against netapp-qtrees-v2-64-bit-cdot.xml to see what was changed)
- cacti_host_template_netapp-64bit-cdot.xml (import it using cacti)

resulting host template will work for both 7-mode and CDOT. tested on 8.2, but should work on any recent and near future onTap reloease.

ATTENTION - it's a quick hack, nowhere near complete solution, but it's enough to get qtree usage graphs working.

[aftershocks]

Hi vgrig,

Thanks so much for your CDOT templates.. I've been searching for someone who has made them... By any chance, do you have any templates that allow us to see the "percent usage" on volumes?

[aftershocks]

Oh by the way, when you add the device in cacti, are you adding the cluster IP? or the seperate nodes? Because the separate nodes doesn't work for me...

[vgrig]

No, sorry - just took a quick look at NetApp MIB: no usage in percent. But you can create a graph template using KB and some arithmetic - just don't ask me how to do it.

If you want to just monitor volumes and q-trees - use cluster management interface (make sure SNMP is enabled).
However - these templates don't index all the cpus on all of cluster nodes (anyone wants to hack the files btw?), so if you want to monitor CPU busy time you'll need to add another device in cacti and use node management interface. Thank works (i have all 4 nodes in my cluster as separate devices in cacti) as long as SNMP is allowed on that interface (check CDOT docs).

Any questions - let me know. Just keep in mind - monitoring is about 10% of my job and NetApp monitoring is 10% of that, so - don't expect quick replies.

Good luck.

[aftershocks]

Hi vgrig,

Thanks so much for your reply...

When you mentioned "(i have all 4 nodes in my cluster as separate devices in cacti) as long as SNMP is allowed on that interface (check CDOT docs)." How were you able to enable SNMP on nodes? Because from reading https://kb.netapp.com/library/CUSTOMER/ ... b60714.pdf, it mentioned this: “Note: SNMP protocol can be enabled or disabled only cluster-wide, SNMP to an individual node in a cluster is not possible in C-Mode DOT 8.1“

Is it enabled on 8.2 and higher?

I'm not too interested in monitoring CPU... More so the volume usage... If I apply these templates that you have created, would I be able to see the volume usage? I assume so based on your screen shots you've attached. Please correct me if I'm wrong...

By the way, great work! I'm sure it will help others like us out who has NetApp in their environment...

[vgrig]

Don't have any 8.1 CDOT, so can't help you there.

For 8.2, yes, you can enable/disable SNMP only per cluster, but if i remember correctly it enables access to nodes using SNMP as well. It's possible networking on individual nodes is not configured with default gateway? Try snmpwalk from machine on the same subnet.

Anyway, if you only want storage usage - cluster management IP should be enough. And yes, you can create graphs for both volume usage and q-tree... And volume snapshots... And aggregates... And aggregates' snapshots.

Glad i can help - pass it forward if you can.

[aftershocks]

Hi vgrig,

Thanks so much for your help.. Is there a way to monitor the Storage Virtual Machines? Because that's where the volumes are.... Do I have to connect via the cluster management IP and use your templates to poll? Is that possible?

[aftershocks]

so I just tried connecting to the cluster management IP and it works fine... I can connect.. But there are no results when I use your template to query... what am I doing wrong here? I'm also running 8.2..

[vgrig]

define "connecting"? to test if snmp answers on the IP address you're using (i'm assuming cluster management) run from the server cacti is installed on (assuming linux here):
snmpwalk -On -c <your_read_only_snmp_community> -v 2c <your_cluster_man_IP> - there should be answer, including CDOT version.

i just did a fresh install of cacti and it work. here is step-by-step:
1. install and config cacti
2. put netapp_interfaces.xml, netapp-volumes.xml and netapp-qtrees-v2-64-bit-cdot.xml in cacti/resource/snmp_queries and change perms on them to much other files in that dir
3. optional - if running selinux, run restorecon -R on snmp_queries directory
4. restart httpd (not sure it's needed)
5. in cacti interface click "Import Templates" and import cacti_host_template_netapp-64bit-cdot.xml
6. click "Devices" and create you CDOT filer (with Cluster man IP, SNMP community and "Host Template" as NetApp-64bit-CDOT
7. after it creates it, click on "Create Graphs for this Host" - all your volumes and q-trees should show up at this point.

[aftershocks]

Hi vgrig,

So sorry! It's working now... I followed what you said and i'm now able to see the volumes! =D

Now I just need to know how I can "convert" to percent usage so that it can shoot out an alert in Threshold..

Do you have something like that setup in your environment? How do you monitor volume usage levels?

Thanks so much for the templates.. Really appreciate your work..

[vgrig]

In my environment i just monitor space in KB/MB, etc. I use thold cacti plugin and set warning limits on vol/qtree usage. Granted this does not scale well, but works for me.
You can play with cacti CDEFs to create graph template for usage in percent: they actually give it as example "Example 19-5. Disk Usage as a Percentage" for regular cacti disk usage template (http://www.cacti.net/downloads/docs/html/cdefs.html). You can just do the same for NetApp. I may play with it a bit if i get some free time, but no promises.

[aftershocks]

Ok thank you so much vgrig! You've been a great help for this!

[aftershocks]

Hi vgrig,

Here's a quick update... I was able to get v8.3 working but somehow on 8.2, when it tries to query, no volumes come up... Strange... and I'm connecting directly to the node management address..

Anyways, on v8.3 where the volumes are able to be discovered, I created the CDEF based on the link you provided for reference..

However, when I applied it to the threshold, I noticed the value that is picked up by threshold is half the percentage...

cdef=CURRENT_DATA_SOURCE,ALL_DATA_SOURCES_NODUPS,/,100,*.. I verified with a few volumes and they were pretty bang on.. All of them were half the percentage when I manually calculated the percent usage... (total used divided by total space)

So I just added "2,*" to correct the issue and now it's reading the correct percentage...

So my final CDEF is the following: cdef=CURRENT_DATA_SOURCE,ALL_DATA_SOURCES_NODUPS,/,100,*,2,*

Not sure if that's right, but it's giving a more correct reading for me..