#enable analytics and quota reports

1 messages · Page 1 of 1 (latest)

glad skiff
#

@lean ingot @shy river Happy New Year to all.
I'm trying to figure out how to enable file analytics and quota reports in NAbox/Harvest. So far my File Analytics do not appear to be collected. REST is enabled, I think.
Same with Quotas', "No Data".
Can some one help?
Thank you
TasP

lean ingot
glad skiff
lean ingot
#

a quick scan of the FSA technical report implies that ONTAP supports FSA with flexgroups

glad skiff
lean ingot
#

🙂 yes those are not the log files I expected. Those are the nabox-api log files but not the harvest ones. What does dc ps show? Is the nabox-harvest2 container running? If so, try this
dc logs nabox-harvest2 > nabox-harvest2.log ; tar -czvf nabox-harvest2.log.tar.gz nabox-harvest2.log

glad skiff
#

It is 1.1MB compressed. I need to have a second person review it; unfortunately I probably will not be able to send it until Tuesday, we are closed on Monday.

glad skiff
#

Just sent it @lean ingot

glad skiff
#

Lem me know when you get a chance to look

lean ingot
#

will do

lean ingot
#

hi @glad skiff thanks for the logs. There are two clusters that are successfully collecting Rest:VolumeAnalytics: 20 instances and 2 instances. If you check your Prometheus do you see metrics for volume_analytics_dir_bytes_used? Like this

#

The other clusters do not have any instances of Rest:VolumeAnalytics
2024-01-12T19:41:03Z INF collector/collector.go:367 > no instances, entering standby collector=Rest:VolumeAnalytics task=data
2024-01-12T19:41:06Z INF collector/collector.go:367 > no instances, entering standby collector=Rest:VolumeAnalytics task=data

#

One of the pollers also has an auth problem
2024-01-12T16:49:23Z WRN poller/poller.go:678 > init collector-object error="StatusCode: 401, Error: auth failed, Message: 401 Unauthorized, API: /api/cluster?fields=%2A&return_records=true" Poller=**** collector=Rest object=VolumeAnalytics

glad skiff
#

How do I get to that?

lean ingot
#

log into nabox, then edit the URL in the URL bar and change dashboard to prometheus

glad skiff
#

I'm in Dashboards, and I ran the query as above, I see only one cluster, acesg.

#

Do I need to manually add the rest?

lean ingot
#

you typed volume_analytics_dir_bytes_used in Prometheus?

glad skiff
#

Yes sir,t then clicked EXECUTE

lean ingot
#

excellent. If you see data there, you should see data in the ONTAP: File System Analytics (FSA) Grafana dashboard. Can you check there?

glad skiff
#

All panels are populated with "No data"

lean ingot
#

if you click in the Cluster dropdown do you see anything?

glad skiff
#

Ok, I see the cluster and two SVMs I saw in the Prometheous query screen. But how do I add my other cluster?

#

I've already enabled it in ONTAP

lean ingot
#

and if you select the cluster, the panels update with data?
what's the name of the poller for the cluster you expect to see there but don't?

glad skiff
#

'aces' reports no data

lean ingot
#

nabox-harvest2.log:nabox-harvest2 2024-01-12T16:49:23Z WRN poller/poller.go:678 > init collector-object error="StatusCode: 401, Error: auth failed, Message: 401 Unauthorized, API: /api/cluster?fields=%2A&return_records=true" Poller=aces collector=Rest object=VolumeAnalytics

#

password is wrong

glad skiff
#

Isn't it the same password for the cluster monitor?

#

I wonder if the account is not enabled for the app

lean ingot
#

yes that could be, I thought this poller had worked in the past, which would imply a password issue. If it has never worked, yes could be an account issue. Here are the docs about preparing the cluster https://netapp.github.io/harvest/nightly/prepare-cdot-clusters/
you can also do an easy test from the command line with curl like this
curl --user user:pass -snk 'https://10.193.48.11/api/cluster?fields=version'

glad skiff
#

Account is enabled for https, ontapi, ssh

lean ingot
#

replace user, pass, and ip address

glad skiff
#

Am I missing something?

lean ingot
#

those look fine. Try the curl

glad skiff
#

Sorry, where would I find it

lean ingot
glad skiff
#

I bet this needs to be alocal account...

#

It acutally took the domain account

#

Spelling errors, all mine

#

It took the account and password which is the same one I've been using, but it has collected any data yet.

#

ok, I found it

#

I'm not sure what the working cluster is using to login with, I'll track it down

#

New DISA rules, they want all AD domain, but the clusters are in separate domains with no trusts. So I'm using local RO accounts. Did not create those

#

Thank you

glad skiff
#

Just modified volume_nalytics.yaml to 10 records / 100 directories, thinking 10 flexgroups and 100 qtrees? Am I assuming correctly?

vapid hazel
#

@glad skiff The filter section outlined below is configured to collect File System Analytics (FSA) for the top n (in this example, 10) volumes—both FlexVol and FlexGroup—sorted by space.used. The MaxDirectoryCount parameter specifies the number of directories (in this example, 100) to be collected per volume, ordered by directory bytes used on-disk.

  - filter:
      - analytics.state=on
      - order_by=space.used desc
      - max_records=20

plugins:
  - VolumeAnalytics:
      - MaxDirectoryCount: 100

Sample URL which collects FSA per volume
api/storage/volumes/375dd429-802b-11ed-a5cd-00a098d39e12/files?return_records=true&fields=analytics.file_count,analytics.bytes_used,analytics.subdir_count,analytics.by_modified_time.bytes_used,analytics.by_accessed_time.bytes_used&order_by=analytics.bytes_used+desc&type=directory&max_records=100

glad skiff
#

Ok, I'm showing my lack of knowledge... Is there a document where I can also define the rest-roles required with readonly access to give HARVEST full access to the API's? i.e. my Node dashboard is all NO DATA.

#

Is it creating a 'services web access" to the current harvest role?

lean ingot
#

hi @glad skiff did you use the role based access control list defined here? https://netapp.github.io/harvest/nightly/prepare-cdot-clusters/#ontap-cli We found a few missing permissions yesterday and that list has been updated. Can you copy/paste those again and restart your poller to make sure you have them all?

If there are permissions problems, they will be present in your logs. I know it's a chore for you to send us your log files so I can give you commands to run on nabox if you like, otherwise please send us your nabox log file.

glad skiff
#

Great, thank you

#

I did a role with -cmddir DEFAULT and -access readonly. Is that okay? Now I will add the REST portion

lean ingot
glad skiff
#

That does not work for some reason. I had to add the web access before HARVEST was able to get data.

#

However, I am seeing that the web-access for harvest2-role has Type admin access. Can I make it readonly?

lean ingot
#

do you mean one of these steps?

glad skiff
#

Yes sir, I believe that is where they came from

#

I also have security and sysmgr assigned to harvest2-role

lean ingot
#

those steps are not using type admin access. the $ADMIN_VSERVER is mentioned because that's how you grant access to ZAPI/REST