Hi All / Ritchie,
I’ve upgraded one of our systems from official 2.8.157 to nightly build 2.8.401.
Looks like all of the data is coming into the Manager correct, I can still see live data, all the input adapters working correctly. I was wondering whether I had to perform a database schema update and data migration to move between these version?
The reason why I ask is that now in Grafana (v9.5.1 old version was v7.3.4), when I explore data, even something simple like querying the OHDATA with a query like ‘FILTER ActiveMeasurements WHERE PointTag LIKE ‘%FREQ’ ORDER BY PointTag ASC’. It is returning all of the correct pointtags, but ‘Field 2’ is blank, where I would have expected the frequency value to be.
Any idea what I’m doing wrong here? Sorry it’s amateur hour.
New Version
Old Version
Cheers,
Scott
I think I’ve resolved my PPA database, I did the schema upgrade and migration, was relatively painless. I don’t think a lot of the grafana dashboards created in 7.x translate nicely to the 9.5.x dashboards, hence some of my issue as well.
One thing I can’t get my head around though is the connection to the OHSTAT database. When testing the datasource, the connection does not work anymore (it did in the old version fine with the same parameters).
I’ve gone through and compared the grafana custom.ini files pre and post upgrade, can’t find any major differences there.
The openHistorian console when testing the OHSTAT connection…
![log](//discussions.gridprotectionalliance.org/uploads/default/original/2X/0/09385d73b9be4bdee1d022751dfa3a1e80fa591f.png)
The grafana logs are saying…
logger=data-proxy-log userId=3 orgId=1 uname=NMSAD\ohuser1 path=/api/datasources/proxy/uid/000000002/query remote_addr=127.0.0.1 referer=“http://localhost:8180/grafana/d/000000002/system-status?orgId=1&refresh=5s” t=2024-07-25T07:32:12.4849727+10:00 level=error msg=“Proxy request failed” err=“dial tcp 127.0.0.1:6356: connectex: No connection could be made because the target machine actively refused it.”
I did a fresh nightly build install in a VM machine running a SQL server and the http source address for the OHSTAT was “…/instance/1.0-STAT/grafana” in that machine. If I use that address in the operational machine it appears to work somewhat, although not sure whether it is correct.
The difference is in the Grafana openHistorian data source - in the nightly build, this is an entirely new version and we haven’t yet deployed an upgrade script.
For the moment, you may need to completely start from scratch with your grafana.db
. Here is the new blank one: https://gridprotectionalliance.org/NightlyBuilds/openHistorian/Grafana/grafana.zip - note this should be placed in C:\Program Files\openHistorian\Grafana\data
folder. Make sure to backup existing file before replacing it with a new one.
Alternately you can try to “re-create” your data sources in your existing Grafana database, like OHDATA
from scratch - but you might have trouble deleting the old one. Also, all the panels on your existing Grafana screens will need to be updated to the new data source.
And yes, the OHSTAT data source url format has changed as well to accommodate the new data source.
The new data source and backend is new, fast and improved - but there’s no migration script yet…
Thanks,
Ritchie
Thanks Ritchie, that makes sense.
I’ve put the updated grafana.db file on and it appears to all be querying correctly now, thank you.
Just one more question, does this break the OpenHistorianAlarmPanel? I haven’t been able to get that one going on the new version? Is there anything I have to modify in the ALARMPANELSTATE action adapter, or does it require an updated grafanaadapters file for the new version of grafana? Not a huge drama if I can’t get it going though…
Cheers,
Scott
Not 100% sure - I’ll need to test the panel - will add this to the list for testing…
The ‘statushistory’ visualisation in the newer version of grafana is quite useful in tracking the health of the PMU inputs over time. Done this by querying the value of the status flags and setting some different thresholds.
Thanks for the help Ritchie, much appreciated.
Scott