NetApp Harvest ver 1.4

[NetAppHarvests]

Hi Team,

Do we know when will harvest 1.4 available for download and does harvest 1.4 support ONTAP 9.3 ?

[gmey]

Hi all,

do not forget to create a copy of the default template file:

/opt/netapp-harvest/template/default/cdot-9.3.0 --> cdot-9.4.0

We do run this since 3 months with X-codes and now with 9.4RC1

Cheers

Georg

[gmey]

Hi all,

The User-Guide was uodated last year - so September 2017 is the right date for v1.4

Anything else should be like before!

Georg

[muzzafuzza]

WHoo Hoo!!!! All I have to say is this is one of the best tools NetApp offers, If it helps show my appreciation. Every time I use it every customer is like WOW! 

Chris/NetApp please keep up the development and really really love this tool as well as our customers. It has become a Staple for our environment and has made our NetApp experiance truly good.

[TMACMD]

Hey Guys. Love Harvest. Found some "install" issues following the install guides

First, the precursor: Quick Start: Installing Graphite & Grafana

(I ran through CentOS7)

Not 100% needed, but additionally installed python-ldap

It is no longer recommended to use version 0.9.13-pre1 (too many bugs/issues)

0.9.15 should be used for graphite-web , carbon & whisper

It is referenced to "chmod 755 /etc/init.d/carbon-cache".

When doing the offline method, this file does not exist. It needs to be copied from the /tmp/graphite_local/carbon-0.9.15/distro/redhat/init.d location first

For the grafana install, referring to their web page, this is what needs to be done

  wget https://s3-us-west-2.amazonaws.com/grafana-releases/release/grafana-4.6.3-1.x86_64.rpm
  yum install initscripts fontconfig (can be added to the list of rpms in the beginning)

  rpm -Uvh grafana-4.6.3-1.x86_64.rpm

And lastly for the NetApp Harvest Install & Admin guide 1.4, a couple of details:

If you are using the offline method, you would need to go to a live system and at a minimum download perl-Excel-Writer-XLSX

(plus possible other dependancies, depending on currently loaded RPMs)

  yumdownloader perl-Excel-Writer-XLSX

That must be installed for harvest1.4 to install

For noobs, you might want to indicate that the supplied file from the NetApp Harvest site must be extracted first

  unzip netapp-harvest-1.4.zip

Then you can do the 

  yum install -y  netapp-harvest-1.4-1.noarch.rpm

That's it for now.

[LDTGT]

Hi, Anyone know if 1.4 is compatible with OCUM 7.3 or are we now waiting for Harvest 1.5 before we can update?

Thanks

[yannb]

No it is not tested with 7.3, but it doesn't mean it doesn't work.

If you are using NAbox, you can just force compatibility and see if there is any problem, if not, you can just go in the templates directory and copy the 7.2 template as 7.3 and see if it works.

[AlexV]

Just installed version 1.4 but it's not working correctly with OCUM 7.3:

[aggregate] update failed with reason: No row with the given identifier exists: [com.netapp.dfm.entity.inventory.ontap.fas.Cluster#67017]
[aggregate] data-list update failed.

[volume] Cluster name for aggr [fas_c1:fas_aggr1] (3eeb4418-8593-11e5-8347-12ab986bb745:type=aggregate,uuid=62ff4490-dfe4-447a-a2e4-db64585d114c) not found in cache; skipping

Do we need to modify the template for OCUM to get this working or is this something in the harvest-worker script?

Regards,

Alexander

[yannb]

Hi Alex,

Where does that error come from?

It looks like a problem with OCUM more than Harvest, is OCUM correctly collecting data about that cluster/aggregate?

[gmey]

Hi all,

did it work previously with OUCM 7.2 ?

And did you create a copy of ocum-7.2.0.conf in /opt/netapp-harvest/template/default

Can you send me the complete log of this worker located in /opt/netapp-harvest/log

With a coipy of the conbf file - you should see this entry in the logfile:

[2018-02-15 15:01:40] [NORMAL ] [main] Found best-fit monitoring template (same generation and major release, minor same or less): [ocum-7.3.0.conf]

I will add this default conf for OCUM 7.3 asap 

Cheers

Georg

[JasonNetAppCCT]

I am working on a CCT escalation for Samsung and they recently upgraded to 9.4RC1. I am confirming if they installed the SDK for Harvet v1.4 afterwards.

One issue that is seen in 9.1 and 9.3P4 is that Total Avg Latency appears to be incorrect when monitoring the system in Grafana. See the 1459.4ms in the screenshot attached.

The latency is correct from CLI and in OCUM, but Samsung still wants to use it for monitoring and fix this cosmetic issue.

Other granular counters are correct – like NFS ops – other_ops – readdir.

Any idea what should be looked at from the system logs to correct this?

Jason Smith - NetApp CCT

[gmey]

Hi Jason,

which version of Harvest are you actually using ? v1.4 ?

The metric in this particular dashboard is a calculated latency from the Harvest volume plugin under the tree:

../netapp/perf/<site>/<cluster>/svm/<svm>/vol_summary/avg_latency

Can you use the Graphite UI and go along the tree and see if this high values are shown all the time ? 

I did just check our test systems and I do not see any SVM with strange "average" values.

This is the code generating these summary counters ... so you may check total_ops for the volumes ... may be there is a problem

# Add per SVM counters
if ($emit_level{svm})
{
#Finish math for weighted averages
$tmp_h{read_latency} = $tmp_h{read_latency} / $tmp_h{read_ops} if ( (exists $tmp_h{read_ops}) && (exists $tmp_h{read_latency}) );
$tmp_h{write_latency} = $tmp_h{write_latency} / $tmp_h{write_ops} if ( (exists $tmp_h{write_ops}) && (exists $tmp_h{write_latency}) );
$tmp_h{other_latency} = $tmp_h{other_latency} / $tmp_h{other_ops} if ( (exists $tmp_h{other_ops}) && (exists $tmp_h{other_latency}) );
$tmp_h{avg_latency} = $tmp_h{avg_latency} / $tmp_h{total_ops} if ( (exists $tmp_h{total_ops}) && (exists $tmp_h{avg_latency}) );

foreach my $counter (keys %tmp_h)
{
push @n_emit_items, "$prefix.svm.$svm.vol_summary.$counter $tmp_h{$counter} $timestamp";
}
}

[JasonNetAppCCT]

Thanks Georg - Here are the versions

Data ONTAP 9.4RC1

NAbox 2.4.2

NMSDK 9.4

Havest 1.4

Will check with Samsung today on the tree information you listed today - Thanks Jason  

[JasonNetAppCCT]

Appears the Graphite UI shows the same abnormally hig latency but the individual counters are within range.