So I’ve been confused why does my HAOS installation use random bursts of bandwidth with a set of intervals. I’ve been disabling and enabling devices, but nothing seems to have an effect.
I do not have cameras or speedtest running and if you look at the picture then neither of those make sense. I do not have external backup running either, only local once a day.
Over time the amounts of data in the burst increases. This is a clean install that was restored from a backup (I can’t really lose all the configs), but the data bursts started small on a new installation from the recovery.
In the previous install they were already around 2.15GB a burst. Right now they started around 10MB and reached over 70MB in 24H.
These bursts seem to be local, only to do with HAOS machine, I can’t see any of those reach out in the internet so the problem is somewhere within the OS. Could there be some sort of leaking docker network somehow?
First time I noticed these was around end of January, I’ve been disabling and enabling all possible integrations and code changes I can from that time besides rolling back Home Assistant.
Any ideas how to solve this as over time it will start to look really ugly on network stats?
It does show up on the network stats for that specific host. And it does show that data on the summary page, while not messing up entire data then for this host any data is useless and it is top consumer in every table.
I do not understand why such traffic should go even outside the host if it is internal DB. Also, this was not present for years, only a thing that started this year.
DB size is around 1.2GB and oldest entry is Feb 17th. As I did migration yesterday then I would say 2.15GB bursts day before make even less sense. Size of the database is not an issue, more of those bursts happening over the regular network, not inside the host.
My guess is that it is a database transaction, like a statistic sensor or likewise.
In your picture you can see it is SSL/TLS traffic, which indicates you might be running with SSL on your HA installation.
SSL is only working with domain names, so it can not work with the internal docker network only, because there is no way to look up names on those IPs.
This means all the traffic has to make a trip around the public network interface, which probably what you are seeing here.
The traffic is not getting out on the public network, because it will be handled internally on the network stack on the interface.
I totally agree with you here about the traffic not making it to the public network, but it is annoying nevertheless since it did not exist previously.
I did try to remove the URL configured in HA, but that did not yield any results. I’ll keep playing around and see what can I disable/enable to get this solved
