[NetApp Harvest] API get different instances count

[koza]

Hello,I have a questions regarding Harvest.

I experinced that NetApp cluster annswer different the number of instances(volume) in each case when Harvest send API requests to a NetApp cluster.
e.g. 1st try : 2919 instances

         2nd try : 3453 instances

I used <perf-object-get-instances> as API.
I would like to know why this API different the instances in each case.
Please tell me the root cause and solution ?

※debug logs
before service restart
[Tue Apr 23 18:40:01.010 2019] [2019-04-23 18:40:00] [DEBUG ] [volume] Starting new poll data batch [0 to 499 of 2919], batch size is [500]

after service restart
[Tue Apr 23 19:00:01.413 2019] [2019-04-23 19:00:00] [DEBUG ] [volume] Starting new poll data batch [0 to 499 of 3453], batch size is [500]

Thanks,

[aattar]

Hi @koza 

Is Harvest directly talking to cluster? or it is talking to OCUM and fetching the data?

[koza]

Harvest is directly talking to cluster.
thanks.

[vachagan_gratian]

For performance counters, Harvest is talking to the cluster itself. For the capacity counters it talks to OCUM.

[vachagan_gratian]

Hey Koza,

This looks quite strange! But I think the root should be not Harvest, but the ONTAP API. What I would do is check the number of volumes using the tool perf-counters-utility that comes to Harvest. Here's how you can do that (assuming that you're in the netapp-harvest directory):

./util/perf-counters-utility -host HOSTNAME -user USERNAME -pass PASSWORD -in -f volume | tail -n +2 | wc -l

The tool will give you a list of volume instances on your cluster and wc -l will give you the total count.

[koza]

Hi @vachagan_gratian

Thank you for your cooperation.
I executed your sugettion.
As a result,utility got the following value.

1st try
[root@XXXX util]# ./perf-counters-utility -host XX.XXX.XX.XXX -user XXXX -pass XXXXXX -in -f volume | tail -n +2 | wc -l
3498

2nd try
[root@XXXX util]# ./perf-counters-utility -host XX.XXX.XX.XXX -user XXXX -pass XXXXXX -in -f volume | tail -n +2 | wc -l
3500

3rd try
[root@XXXX util]# ./perf-counters-utility -host XX.XXX.XX.XXX -user XXXX -pass XXXXXX -in -f volume | tail -n +2 | wc -l
2967

[bkamil]

Isn't is the case that only the instances that are in use (IOs happening on them) at a given point in time get reported?
I think I've seen similar behavior with AV scanning statistics.