I noticed since the gallery changed up with this latest update various posts are taking quite a bit longer to load than normal. especially videos. and considering how fast my internet is videos taking that long to load when they usually load up in a second or two is a bit odd.

Might not be applicable, given that it's a recent issue for you, but the suggestion in topic #56792 worked for me when I was having slowdown specifically on this site alone.

kathyohneke said:
I noticed since the gallery changed up with this latest update various posts are taking quite a bit longer to load than normal. especially videos. and considering how fast my internet is videos taking that long to load when they usually load up in a second or two is a bit odd.

it because Cinder decided to increase all the thumbnails sizes which take longer to download and e6 is regenerating thumbnails so the servers are already under extra load

funkwolfie said:
it because Cinder decided to increase all the thumbnails sizes which take longer to download and e6 is regenerating thumbnails so the servers are already under extra load

I'm not sure why you feel the need to call him out specifically, it was also already said in the discord that the server doing the sampling was taken out of rotation so the lag should not be caused by that

cinder said (discord, screenshot):
The slowdowns should have been largely addressed, I believe?
The reason for those was that the server that is processing thumbnails had also been used to serve content to users. Normally, that isn't an issue – but because it's currently bogged down with a few million images to generate, it caused lag.
The issue was inconsistent because requests wouldn't always go to that specific server.
Now that it's been taken out of rotation, the lag spikes should subside.

donovan_dmc said:
it was also already said in the discord that the server doing the sampling was taken out of rotation so the lag should not be caused by that

Just to clarify, there was also a secondary incident there, as scheduling several million jobs caused redis to run out of memory.
But that should have also been addressed yesterday, with sampling jobs being passed on to sidekiq at a more reasonable rate.

cinder said:
Just to clarify, there was also a secondary incident there, as scheduling several million jobs caused redis to run out of memory.
But that should have also been addressed yesterday, with sampling jobs being passed on to sidekiq at a more reasonable rate.

I can confirm that server stability and speeds have gone back to normal
The TagMe site doesn't scream api "the e621 servers have burned down" failure at me every two submissions for tag and set changes

nin10dope said:
I can confirm that server stability and speeds have gone back to normal
The TagMe site doesn't scream api "the e621 servers have burned down" failure at me every two submissions for tag and set changes

The server fires had been put out too.

post #2451564