search

opendata-stuttgart / madavi-api

api for rrdtool graphics per sensors + graphics from archive.luftdaten.info
19 stars 11 forks source link

Missing data on Madavi graph/csv (but not on http://maps.luftdaten.info and https://opensensemap.org) #13

Open MagTun opened 5 years ago

MagTun commented 5 years ago

I checked on several others sensor and they all missed data in the past 10h.

madavibug

Something wrong with your server?

ricki-z commented 5 years ago

There were too many people looking for the new years graphs. So there were some performance issues.

MagTun commented 5 years ago

We are now January 6th and it's even worse (my measuring interval is 145 sec). github

My graphs on luftdaten and opensensemap don't miss data like the one on madavi. Even connecting to the link https://www.madavi.de/sensor/graph.php takes a lot of time and sometimes I get the error "www.madavi.de is currently unreachable." or "www.madavi.de took too long to respond."

If you need to upgrade your server and need money for that, please post a link on your website for donations.

WiliTest commented 5 years ago

Same problem.

thjean commented 5 years ago

It is striking that while the csv-Files on madavi contain all posted values from several days, the graphs are ok for one of my sensors, while they are completey fragmented or even empty for the other sensor.

As this is reproducible at different times of day and because of the complete csv-files, I tend to question the new-year-story.

ricki-z commented 5 years ago

@thjean measurement interval is 145 seconds on both sensors?

thjean commented 5 years ago

This is an interesting hint. Actually the sensor with the complete graphs has a measurement interval of 180 seconds (this sensor is running airrohr fw) while the other sensor (own firmware, does not send the value-types "sample", "min_micros" and "max_micros") has a measurement interval of 150 seconds. The interval was at 60 secs some days before and no more data have been shown in the graphs since the interval was changed to 150 seconds.
I will set both to 145 seconds and see if there is any change.

thjean commented 5 years ago

Adapting the measuement interval to 145 seconds did not lead to any improvement. The root-cause for my issue seems to be related to the "software_version" string and does probably not apply to the OP's issue.

As long as the sw-version string does not begin with "NRZ-YYYY", the graphs are not displayed. Prepending "NRZ-2019" to the individual software-version works for my sensor and the graphs are shown.

[Edit] The software-version string must not start with "NRZ-YYYY", it just needs to be >= "NRZ-YYYY...." in order to display the graphs. Sending "Vxyz" as software-version string works. Referring to #10, it may be required to set the measurement interval to 145 seconds for the display of all graphs. [/Edit]

Looking at the graphs at https://www.madavi.de/sensor/graph.php suggests that this is not just my issue, as there are more then just a few sensors (with sensor names leading to the assumption that they do not run the airrohr-fw) showing empty graphs as well. I'm just wondering why the 24h floating graph showed data as long as I posted with a measurement interval of 60 secs, as this was before adapting the sw-version string.

MagTun commented 5 years ago

My sensors were set to 145 sec even when , so I don't think this is the reason.