I keep posting threads today. So I tried upgrading to 1.0.2. Uploaded the files and went to /upload. After a couple of minute rebuilding I get this error:
An error occurred or the request was stopped.
And then I get to a nginx page which tells me that there was a gateway timeout. All good and well, but then when I try to access the forum it still tells me that there is an upgrade going on, and when I try to access the access panel it tells me to press the upgrade button and it starts all over again. I tried it twice now, what do I do?
EDIT: I editted a timeout in the nginx.conf from 65 to 130. Maybe that helps.
Presumably it was the first step of that process, as the other steps will be split into 10 second chunks. You would've gotten through this process successfully in the past, so unless it was just on the cusp of failing before, you should be able to get through it.
Are you running a VPS? The cache rebuilds are very I/O intensive, and we have seen some VPS's with really shocking write performance. (10% the speed of what I see on my desktop, for example.) It's a tricky problem to solve.
Yes, I'm indeed on a VPS but I doubt IO is the problem. I share the disks with only 3 users, and iotop shows me that mysql is continuesly busy with writing abou 30K/s to the disk, with spikes like 3M/s.
Setting the timeouts to a higer value appears not to work. I switched back to apache just now for just the install, let's see if that works.
Apache has the same, BUT has a 'Rebuild Caches' button which continues the process. Yes, it still takes too long pn apache so it times out, but I just look at iotop until the mysqld isn't busy anymore and push that button. Nginx appearently just shows an timeout and refuses to go on.
So this makes it a nginx-related problem for going to it's own page instead of showing that button?
It managed to do the upgrade with apache just fine, and after it I switched back to my beloved nginx.
Actually, it's supposed to automatically go to the next page (with a non-AJAX request, which gives it more chance of working), so I'm not sure why that didn't happen.
I guess with nginx, it killed PHP after a certain amount of time, whereas with Apache it ran to completion. It really shouldn't be taking that long, even with 2 languages. I'm not sure if I can find a way to improve the speed.