The dashboard "ONTAP: StorageGrid FabricPool" that was introduced in the 23.05 release does not list all the clusters in the list of "ONTAP clusters" on grafana. It only displays a few among several datacentres and clusters defined in the harvest.yml file that every other dashboard is able to correctly display. On trying to investigate this further, it could be noticed that the "Datacenter" tab that is displayed with other dashboards does not get displayed at all.
#Dashboard "ONTAP: StorageGrid FabricPool" has no "Datacenter"
1 messages · Page 1 of 1 (latest)
Hi Madaan - that dashboard is filtering the cluster variable to only those clusters that have a cloud store associated with them. Perhaps the clusters that are missing from the list don't have a cloud store associated with them?
Hi Chris - I now have several clusters that have object store configured. I have checked this with the commands "storage aggregate object-store show" and "storage aggregate object-store config show"
which version of Harvest are you using and which collector? - after 23.08, mamoep discovered a few bugs with collecting these via ZAPI (they work fine via REST). https://github.com/NetApp/harvest/issues/2344 Those bugs are fixed in nightly if you want to give that a try
ok thanks Chris. I will give this a try. If it interests you, I noticed that the clusters displayed were the GCP CVOs with a cloud storage bucket, but none of the on-prem clusters with storage grid buckets appeared. I will let you know how this goes after the upgrade.
thanks! Please do
23.05 was completely broken with ZAPI collector. 23.08 only has minor issues left, but clusters should be available in the list.
Thanks Mamoep !
Hi Chris, I have upgraded the Harvest to the suggested release but the problem still remains !
Hi @olive vessel if you go to Prometheus, do you see the cloud_target_labels metric?
Hi @swift oak yes I do see "cloud_target_labels" and "cloud_target_used" in metric explorers
excellent, that's encouraging. Did you load the new dashboard included in nightly? If not, can you import it?
Yes, I have loaded the new dashboard Chris.
( if you meant tha Grfana dashboard with the new release )
i mean the dashboard changes in the nightly you took three days ago - if you open the FabricPool dashboard and click on the gear icon, then variables, then select Bucket, what query do you have there?
I dont have the settings/gear icon - perhaps my limited privileges to the dashboard ?
yes that's possible - are you using nabox or installed some other way?
nabox - standard installation
and after installing nightly, did you reset harvest dashboards?
It might be a completely different installation, Chris, but I see no settings options at all. Just dashboards. I or my colleagues do not see - or know about the settings options that you show in the screen above
yeah looks like nabox disables that option, not sure if that's intentional or not. @bleak parcel do you know why in nabox dashboards don't include the gear icon?
No it should be there, what version of NAbox ?
i can confirm nabox version NAbox 3.2b (2022-12-04) - Alpine Linux 3.14.2 is missing the gear icon, e.g.
What version are you using @olive vessel ?
Yes, that is what we do Yam - we have enabled anonymous access
Ok, shouldn't be a concern in your case, but the only explanation I see if you don't have the reset button is you have an old NAbox version, in that case, there is a good chance the dashboard import would fail with recent Harvest versions if I remember correctly. You would see that in nabox-harvest2 logs
ok, what would you suggest as the next step Yann ? Look at the logs, and then may be consider an upgarde ? What is the logs directory where I can see this information.
from ssh, you can run dc restart nabox-harvest2; dc logs -f --tail 100 nabox-harvest2
cant say why, but the dc commands dont work Yann
Each of my clusters is a docker instance and this is what i do to restart them ( which I have done several times today )
docker-compose -f prom-stack.yml -f harvest-compose.yml up -d --remove-orphans
You're probably not logged in as root
oh wait
that's not nabox ?
I'm confused, you shouldn't be running those commands or your own compose files
yes, looks like you might be using the docker-compose workflow instead of nabox @olive vessel If so, can you confirm that you upgraded your containers via https://netapp.github.io/harvest/23.08/install/containers/#upgrade-harvest
hi Chris, yes - that is how I have upgraded the containers. I have just done a docker pull and another restart - to be sure.
but still no gear icon in the dashboard?
So Chris - a very interesting observation. I now have more clusters in the "Ontap Clusters" drop down. I only had the GCP clusters before the upgrade - but I now have all the on-prem clusters where I have enabled FabricPools tiering
just that the "Datacentres" drop down is not there - and of course no gear icon, or the other settings option that you had showed me earlier
right, there won't be a datacenter variable on that dashboard. The most recent version should look something like this
Can you try logging into Grafana with the default credentials? https://netapp.github.io/harvest/23.08/install/containers/#grafana
ah thanks - If we only expected more clusters and not the datacenter variable from the new release - then I think we have that covered now 🙂 I had just been looking for that variable to appear without realizing that the new clusters have been there already
perfect!
Just logged in with default credentials and I see more options now - let me give you the information from that
Just in case you still wanted to see this Chris
thanks! that means you're using an older version of the image that does not include this recent change. Can you copy/paste one of your poller blocks from your harvest-compose.yml file? Here's one of mine as an example:
thanks Madaan - I see the problem. You are using image: ghcr.io/netapp/harvest:latest instead of nightly (image: ghcr.io/netapp/harvest:nightly)
latest will point to 23.08.
The changes we are discussing were made after 23.08 and only in nightly. If you want to try them you can you follow these upgrade directions, except for step one use docker pull ghcr.io/netapp/harvest:nightly instead of what's listed there https://netapp.github.io/harvest/nightly/install/containers/#upgrade-harvest
at step 3, use this instead of what is listed on the page
docker run --rm \ --entrypoint "bin/harvest" \ --volume "$(pwd):/opt/temp" \ --volume "$(pwd)/harvest.yml:/opt/harvest/harvest.yml" \ ghcr.io/netapp/harvest:nightly \ generate docker full \ --output harvest-compose.yml \ --image ghcr.io/netapp/harvest:nightly
Hi @swift oak , I need a suggestion from you. Now that we have upgraded the on-premise instance, I wanted to upgrade the cloud instance too. We have built the cloud instance to monitor the FSX instances which the on-premise instance could not monitor and still cannot monitor ( I just tested this).
Before moving on to upgrade the cloud instance with the nightly 23/08 release, I reminded myself that I had built this instance from a cloudformation template - https://docs.aws.amazon.com/fsx/latest/ONTAPGuide/monitoring-harvest-grafana.html
You can use standard NetApp monitoring tools to monitor your file system storage usage and performance, with the following Harvest and Grafana solution being one example.
and therefore felt that it is best for AWS to release the new template with the new release/patches than update it manually. Would you or NetApp have any recommendation on that ?
although I must admit that the template does not indicate what version we are on. I had to assume that it was 23.05 since I installed it 4 months ago.
hi @olive vessel I believe @quiet cosmos was working on some updated documentation to cover upgrade
@olive vessel this hasn't gone through the QA process yet, so please make a backup of your environment before you start! but you should be able to do this:
/usr/local/bin/ansible-playbook /home/ec2-user/harvest_install/manage_harvest.yml --tags upgrade
and then
/usr/local/bin/ansible-playbook /home/ec2-user/harvest_install/manage_harvest.yml --tags dashboard
if you're really curious, run docker ps to list your container IDs and then check the version of docker in each container with docker exec $CONTAINER_ID bin/harvest -v
you can do this before & after the steps above to see what changed.
thanks Chris!
thank you Chris 🙂
Many thanks @swift oak and @quiet cosmos . I have reached a good postition overall with the upgrade. However, on checking this morning, I noticed that while all my on-premise clusters are reporting OK, the GCP clusters are not reporting since the upgrade. These clusters have been reporting well before the upgrade and nothing on them has changed. Do you know of any bug or issues with the reporting of GCP clusters with 23.08-1 ?
I'm not aware of anything, but I have no experience with GCP clusters so I can't help there
@olive vessel can you grab the logs for one of your GCP pollers and email them to ng-harvest-files@netapp.com?
Because the dockers instances for them are not up, I do not see these clusters reporting/logging on /var/log/harvest. Would there be another place to find those logs @swift oak ?
you can try https://netapp.github.io/harvest/nightly/help/log-collection/#docker-container but if the containers are in a restart loop there may not be a lot there. Worth checking and sharing your harvest-compose.yml for one of these failing pollers too
just sent the logs, Chris
received, will take a look
Phew..I reached the resolution Chris. The user and its creadentials were only added to the clusters with the "http" application - but I have now added them with the "ontapi" application as well 🙂