#ONTAP: FlexCache

1 messages · Page 1 of 1 (latest)

peak ferry
#

How do we get this report properly working? Right now only FlexCache details shows anything, how can we get the something like the "Top 50 Cache Miss percent" window populated. I'm running the latest 25.11.0. What do I need to change on the NAbox server?

austere tulip
peak ferry
#

Uploaded

#

thanks

#

Please don't add any sensitive information here on this thread from the logs 🙂

austere tulip
#

thanks for the upload. No worries, we won't

#

it looks like most of your clusters don't have FlexCache instances, but some do. For one of the clusters that does (I can give you the poller name of one if you want), can you edit your /etc/nabox/harvest/harvest.yml and change the collectors to:

collectors:
    - Rest
    - ZapiPerf
    - RestPerf
    - Zapi
``` instead of what you have right now? Save the file and restart
peak ferry
#

I will make the change and update.

#

You want me to leave Ems or remove it?

austere tulip
#

fine to leave

#

it was not included in the poller I looked at so I did not include in the list above, but it's fine to leave

peak ferry
#

Interesting. I changed a few and now lost access to the GUI for Nabox and grafana.

austere tulip
#

perhaps you made an error in editing the file? Maybe used tabs instead of spaces?

peak ferry
#

Well this is not good.

austere tulip
#

we can fix it

peak ferry
#

Hmmm

austere tulip
#

what happens if you run this after sshing into nabox
docker exec havrest bin/harvest doctor

peak ferry
#

Error response from daemon: No such contanier: havrest

austere tulip
#

did you use vim to edit /etc/nabox/harvest/harvest.yml?

peak ferry
#

vi

austere tulip
#

ok let's open the file again in vi. After you open it, let's search for any tab characters by typing
/\t then Enter

#

you can also tell vi to show you the tabs by typing
:list
tabs will be shown with a ^I like so

peak ferry
#

Yeah I don't have any of that in the file.

austere tulip
#

and the search for tabs returned nothing?

peak ferry
#

correct

austere tulip
#

did you change one poller or several?

peak ferry
#

Yes, but changed them back.

austere tulip
#

ok with your reverted changes, does the container start?

peak ferry
#

Negative

austere tulip
#

what about
docker logs havrest --since 1h

peak ferry
#

Same error essentially

#

erro reponse from daemon

austere tulip
#

try

dc down
dc up -d
peak ferry
#

Aww now I"m back

#

So, does this work in your lab?

#

for the FlexCache

austere tulip
#

yes

#

it will work if you use the order I gave you

peak ferry
#

Hmmm

austere tulip
#

the restart problem was caused by a bad edit to the harvest.yml

peak ferry
#

But there is nothing wrong with changing multiple corect?

#

I mean, I will test with one single and see if that works.

austere tulip
#

nothing at all

peak ferry
#

I will try vim for what it's worth since it's newer

austere tulip
#

they're the same - vi is a symlink to vim. If memory serves, running vi will run vim in vi mode when it is symlinked

peak ferry
#

I changed just one item and restarted. I will give it some time I take it to display some data on that report.

austere tulip
#

cool, so everything restarted fine this time?

peak ferry
#

Yes just with changing the setting for one host.

#

I assume I just need to wait a few for the data correct?

austere tulip
#

yes, ~4 minutes. We can also check the log files if you prefer

peak ferry
#

I will give it a few

#

Hmmm I guess it really didn't like when I did multiple at once, who knows. I did another one and it restarted just fine as well.

#

Now let's hope I see some data 🙂

austere tulip
#

indeed

peak ferry
#

No extra data yet.

austere tulip
#

please upload another support bundle at the same location

peak ferry
#

I use rest collectors.

#

ok

#

Before I do that. The "Defaults", I don't need to change do I? I only changed the ones under each array I have with Flexcache.

austere tulip
#

right, the defaults will only be used if a poller does not define its own

peak ferry
#

OK then here it comes

#

Done

austere tulip
#

thanks for the log files, they explain the problems. You changed three clusters - one has the wrong collector Zapiperf it should be ZapiPerf. The other two pollers have the correct poller name, but Zapis are disabled on the cluster so the ZapiPerf collector isn't going to work.

The issue with FlexCache is ONTAP does not support collecting that object via RestPerf. You choices are either a) enable ZAPI for those clusters b) Setup the StatPerf collector to collect FlexCache performance metrics (assuming these clusters are not FSx). If you want to setup the StatPerf collector see https://netapp.github.io/harvest/25.11/prepare-cdot-clusters/#statperf-least-privilege-role

peak ferry
#

Dang REST man. LOL

#

OK I will take a look at those steps. Thank you!