NanoSD super slow, data falling behind, buffer filling up

[mcl1981]

Good morning. I'm having a problem with my Nano SD. This started a few weeks ago and continues today. I thought maybe it was the new version so I downgraded down to 5.1 but didn't change.

First, it has become extremely slow. It takes a very long time to click around and do things in the Web UI. It takes a very long time to even type let alone run commands in an SSH shell. The Windows file shares are also just as slow.

Second, as time ticks on since boot, the buffer fills up more and more and the data it's uploading gets further and further behind. After 12 hours of uptime, the data upload is 10 minutes behind and there's lots of stuff in the buffer.

These problems persist after soft reboots and power cycles.

[wvdkuil]

Often happens with to much retries and to short upload times.
A small, not yet detected change somewhere between your Meteobridge and the sites you are uploading to is all is needed.

So if you set uploads f.i. every 5-10 seconds => do not set retries
If the receiving server is (temporary) to busy the que will grow and the Meteobridge will try to upload all "missed" uploads.

See discussion viewtopic.php?p=34459#p34459

Succes,
Wim

Tons of explanations... the server you transfer to might got slower because of more load or you have different response times in your Internet connection... There are tons of parameters that are important but which are out of your reach and no one will even tell you when something changes there. This is infrastructure not under your control. Therefore, you shouldn't push that to the limits, if you are interested in stability.

[mcl1981]

All services are set for no retries. I tried turning off minute data, which also didn't help.

[wvdkuil]

All services are set for no retries. I tried turning off minute data, which also didn't help.

If you make a change you have to restart the MB also.
Otherwise the 100+ uploads are still in the backlog and MB is still trying to upload them.

If after a restart the backlog grows again, you could switch off half of the uploads ,
=> restart
==> if the backlog remains at 1 or 2 uploads, the problem is one of those you switched off.
==> otherwise switch off the remaining ​

Then add one by one the uploads and check what is happening.

Wim

[mcl1981]

Yes I reboot after all these changes to clear the backlog. Is there some way to see what is in the buffer? I'd like to just see what is holding it up rather than randomly deleting and rebuilding all the services and nets.

[mcl1981]

I changed all my 1 minute or less updates to 5 minutes. Left WU and my own in-house mySQL server at 1 minute. No more buffer.