Tonight (and after a long time only, so actually the first time while using Homeassisstant) our Internet went down for several hours. I found HA almost unusable this morning. The log files was totally messed with access issues (like “Telegram Updater failing”), “invalid HACS Github Token” etc. kind of messages, at some point seemingly even using up available connections altogether. Lots of pages did just open empty screens (HACS, Simple File Editor, Node Red and others).
While it’s clear that in such a case the Google Drive backup fails and the weather report & forecast are unavailable, I don’t get why other functionality (that should basically keep working with my up & running home network) seems to be so susceptible to an Internet outage. I restarted HA, and Grafana, Node Red, and the simple FileEditor don’t even start any longer (which specifically for Node Red leaves a lot of my automations now in limbo).
I am always amazed by the high quality standards applied to all the HA content, but shouldn’t “graceful behaviour while Internet outage” be part of the score card to be ticked off before release of integrations? So e.g. pause update activities for a reasonable time if not working, start properly with local resources & content etc.
These are to be expected, as they are cloud services.
Grafana and FileEditor not, but then I think, do you use outside access using a domain name. If so, did you setup split-DNS ?
But still e.g. the telegram updater way too often retried connection (and there was no message to send also, I just use it for some rare alerting). And I think HACS should still launch for accessing the installed integrations and extensions, even if it cannot access remote repositories for the time being.
I am addressing internally via “homeassistant” local DNS name (which is still properly resolved) or IP only. I am not using it from the outside (or just via VPN, so basically also “internally”).
Can you do something with the logger integration to filter internet-related messages?
But just not seeing them wouldn’t help with the fact that a lot of stuff just doesn’t work, though it probably could (and should…).
What do you need HACS for if you don’t have an internet connection? You can’t manage the stuff you installed through HACS with HACS itself, only install or update, which is anyway not possible without an internet connection.
Grafana should run just fine without internet, at least it’s the case in my system, but I use HA Container not OS. I think the OS has some general problems without internet as it does some checks pretty often, which require internet.
Well, I could still deinstall current plugins, and also this is not the point (and HACS not my major issue at all, rather Node Red runtime). Overall, plugins should IMHO behave graceful and work with minimal functionality that does not require Internet connection. E.g. updating or according polling functionality can well be delayed. If it’s not a cloud service per se, it should work on a standalone HA instance even w/o current Internet access.
Furthermore, HA did not recover even when Internet came back. Only once I did a full host reboot, operation was back to normal.
Totally agree, for me everything that doesn’t require an internet connection continues to work after an outage (I actually tested it by capping the internet connection as I did not have any outage in the last 6 years). So I’d assume it has to do with Home Assistant OS.