I forgot to add that I am running the smallest configuration for this test. One KIWIsdr and only one 20 meter channel. I do see the recorder process in the KIWI.
John

Still working on the same problem to get wsprdaemon to run on the Nvidia Nano running UBUNTU 18.04.
The problem I am seeing is no spots with the tail -F uploads.log command. I do see the message to purge the empty file.
I have kiwi built in wspr app running on the kiwi and I see spots on the same 20 meters/
Rod had me add my CPU type arm64 to the wsprdaemon script. I did not see any change.
I guess that maby the wsprd program that Rob imports from wsjtx may not be compatible with the arm64.
I spent the week building wsjtx from source on the Nano. I then moved the new wsprd into the bin folder.
I ran wsprdaemon again -- no change.I started wsprdaemon and then wd -d three times to get the high verbose.
I let it run for 12 min. I have attached the wsprdaemon log file but I do not know what to look for. Rob is still tied up with the new house and version 3
so I thought I would post here and hope for recomendations.
Thanks in advance for any comments
John

and how many bands are you trying to decode?

what OS are you running?

This error shouldn't appear in more recent Kiwi versions. When the non-NBFM squelch was implemented the API was changed slightly. But there was a period of time where the old API was not accepted as it should have been. The old API is used by directTDoA, kiwirecorder and, it looks like, wsprdaemon.

(Sorry, only seeing this now as I don't check here very often.)

The new (non-NBFM) squelch function only effects the browser. Like the mute button.

But kiwirecorder connections are coming in on a different channel than a browser connection. So there wouldn't be any interaction anyway. There has been a suggestion that a muted channel should stop sending audio to the browser to save network bandwidth. But even if that were implemented it still wouldn't effect kiwirecorder connections.

Kiwirecorder has its own squelch setting that pauses recording to the file to save space.

I have been working to get the daemon running on the 4GB nano. Most is looking well. The KIWIs are loading ok just no spot uploads.
At decode time { 8 sec mark} I see a bump in CPU to 80% but it only lasts for 3 seconds. How can I check on the wsprd processes? Can I learn anything with three -d ? Is there a place in the wsprdaemon.sh  where I might trap an error message?

John
TI4JWC

All fixed. Was ethernet switch problems./

I have found the update takes about 17 minutes. Try and do a full power cycle after 17 minutes

--

I have two Classic KIWI stuck in Update. One from 1.445 to 1.446 and the other is just sitting at 1442 with 1446 Available but with no action. Both are not accepting settings from wsprdaemon.
Any ideas?
John
TI4JWC 

Hello All,

I was going through their examples (on my local workstation), testing things out, and getting a feel the CLI / API. I decided to load one of their bigger datasets and play around some. I chose the Airline OnTime-Dataset first and went through all the example queries. This thing is seriously fast, and that's putting it mildly.

I don't know what it would do, or how it would handle, hundreds of concurrent users, but, if it scales with clustering the way it's performing on a single node, it's hard to imagine using anything else (at the price point it offers) with the anticipated volumes WSPR is expecting over the next cycle.

I realize there is still a fair bit of ETL needed before shoving things a into the warehouse, but, I can see use cases of all sorts with this thing.

73's
Greg, KI7MT

Hello All,

I should have looked a bit further before diving into a complex solution.

Third Party UI Clients are Plentiful

I already have a subscription to JetBrains DataGrip (and others), and it has native support for ClickHouse. All I had to do was look for the connector .. Doooh !!!

In any Case, the more I learn about ClickHouse, the more impressed I become. This thing is powerful, and seriously fast !!

73's
Greg, KI7MT

Disaster recovery / backups are never important until ya need them (been there, done that, lost the data).

I've got a full copy of the WSPRnet dataset also, in multiple locations. Not that I really need to, but, its nice not having to re-download those big files.

I've got data center infrastructure on Azure and AWS in multiple zones also. It's costly to have and maintain, but, it's there when the lights go out in a zone.

We maintain 3 copies of the wsprdaemon.org site:  

1) wsprdameon.org is a cloud server maintained by digitalocean.com on a Silicon Valley server
2) wd1.wsprdaemon.org is a Dell server located in a telco engineering lab in Sacramento California , 200 Km from WD2
3) wd2.wsprdaemon.org is a Dell server located in the Hurricane Electric Tier 1 peering point in Fremont California in Silicon Valley

DNS is handled by Google.

So unless all of California burns up or slides in the Pacific Ocean, I think we have a pretty reliable system

Yeah that was bad luck. I never considered wspr.live to be important enough to include it in my backups and now it is gone into the cloud forever. The dashboards and the information page are lost.

But I just rewrote the page and will put Grafana back on air soon.

The Database itself is easy to restore in case of loss as long as wsprnet.org is available we can just reimport the data.

Hello All,

I'm probably late to the party, as usual, but, I was playing around with Phill's wspr.rocks site, and it definitely Rocks on the speed front !!

I'm looking to explore ClickHouse Database solutions a good bit more and was wondering two things:

• Has anyone published their WSPRnet / WSPRDaemon Clickhouse table schemas / Index recommendations?
• Has anyone used an PostgreSQL Foreign Data Wrapper (FDW) integration to allow query's via PostgreSQL interfaces?
  • clickhousedb_fdw
  • infi.clickhouse_fdw
  • or others

I have a server set up with ClickHouse and walked through a couple of their DataSet examples. There's no question, the thing is seriously fast.

Thanks

73's
Greg, KI7MT

Just saw this today.

Thankfully, the DB looks to be functional still.

Does the threshold control have any effect on kiwirecorder, or is it just for the audio that is heard in the browser?

73
Steve KD2OM