Contain your excitement, Google says it’s next week
No I don’t and can’t. But mobile performance will be slower anyway so it’s more likely to trip over error messages.
Its a bit unmanageable. The thread doesn’t update in real time when it happens so you’re often having to come back out to the main board and then back into the thread. When posting you sometimes get the spinning wheel of death and then the screen goes blank so you’ve no idea if the post went through or not.
When I say well maintained, he uses about 3 applications yet his homescreen is fuckin full of them, yokes I’ve never seen before and it’s an old one at this stage. Slow as fuck and was never put up on a ramp for a check I think.
Worst I’ve seen was during the Dublin Tyrone game last week. Site was borderline unusable for a couple of hours. Be good if people who were using the site yesterday during the hurling could comment on it.
Do you have performance stats/telemetry to show you when the load gets too much?
I did…it was the same…
Yes. But there is nothing extraordinary during those times. It’s more about simultaneous connections, the actual CPU usage and data transfers etc aren’t too bad.
There were definitely some issues with the configuration however that I fixed last night. And there were some issues with the code of the underlying software that have been fixed but are not live here yet.
Just not sure they will make enough of a difference but there are more things I can look at.
what’s the network load like at these times? Someone could be performing DDOS attacks.
Thanks. I’m hoping this is caused at least partly by the site trying to invoke the emergency mode and logging you out so that’s why the reply isn’t working. That is probably not correct but I want to see if reducing the likelihood of the emergency mode kicking in actually improves the experience, even if the server load itself is the same.
I think that’s a different problem. I don’t know of that happening to anyone else.
They could be but they’d be timing them to occur during matches which seems coincidental.
We were definitely underpowered for dealing with simultaneous posting until last night. We had 2 unicorn workers which was not enough for concurrent posting.
Been happening quiet a bit for a few months. It also logs you out , that’s happened me a few times recently.
it’s single threaded as well.
post concurrency is strange, the job is triggered as the post button is actually pressed. Are the posts cached as they are being written? I think they are. That could be an issue.
It seems the message being displayed is the first point of escalation but doesn’t log you out. Its only if you keep trying to do something you can get actually logged out. Interestingly, you don’t need to log back in, if you refresh your browser you get back to main menu logged in as normal.
Not that interesting in fairness
Hmmm you sound like you know what you’re talking about.
Yeah it’s single threaded. I think that’s why the higher number of unicorn workers is required. But I’ve probably gone too high now which will effect the RAM (I think?).
Yeah I think they’re cached. Why are you figuring that’s an issue?
Problem with Mobile on iphone. Desktop was ok. Mainly when there is a big event in progress.
The auto logout thing happens all the time. It doesn’t actually log you out though. It’s a false negative. It says it has, then you’ll get the heavy load message. Then if you refresh you’re back to normal. That’s my experience anyway. And that’s only on mobile(android).