Hacker News new | ask | show | jobs
by Jedd 878 days ago
How would that work? You'd get an in-app message in 40 minutes saying there's a problem 40 minutes ago?

Anyway, this is what status pages are for: https://admin.microsoft.com/servicestatus (currently showing a whole lotta information about problems with Microsoft Teams).
3 comments
cqqxo4zV46cp 878 days ago
Status pages are nerd tools. Anyone that says otherwise is living in a bubble. Pretending momentarily that this is not true, you wouldn’t go to a status page if you thought that everything was fine. Not sure what is warranting such a kneejerk defence…

The implication is that the infra required to notify of an outage is lesser/different than what’s required to…run Teams. Publish it on DNS!

link
Jedd 877 days ago
If you think everything is fine, presumably that's because every metric you have available is indicating fine-ness. That's probably sufficient for most users. There'll be some edge cases, but the brief write-up in some obscure IT online journal should suffice to restore anyone's slightly damaged reputation regarding a torpid non-committal response.

I'd also assume if you're using Microsoft Teams it's because someone else has determined that you shall use it. I expect that any semi-informed, mildly IT literate Microsoft Teams user has a starting position of 'low expectations'.

Anyway, parent was suggesting an in-app message to indicate some issues, so my initial question stands - how do you expediently send an instant message to a user to advise them that your instant messaging app is having a bad day?

link
Log_out_ 877 days ago
This generation will reinvent the watchdog..
link
winrid 878 days ago
Use message sent date for one, and check how late messages are arriving. If late, ping a status check endpoint.
link
Jedd 877 days ago
As a sibling noted, messages on the receivers end are time-stamped with the receive time, not the send time.

To your specific suggestion, how do you identify if the messages arrived late because of a server-side issue, rather than a client-side, network disconnect, or some other non-infra issue?

Anyway, while I'm sure there's myriad ways that the authors of this app could have engineered it to be more resilient, self-reporting, etc - it's clear that they did not.

link
heyoni 877 days ago
That page was green for most of yesterday lol. I stopped checking around 2pm SIX hours after people were noticing problems.
link