I'm using rclone serve http + local disk to create my own http backend.
I have switched two servers today from google drive to my own http backend. And I'm seeing lots of i/o timeout to large files.
Both the rclone serve and the rclone mounts have
--timeout 10m so I don't understand how this can be happening?
The errors starts seconds after restarting my mount, so it's not even waiting 10min?
EDIT: It seems that after changing from domain to IP on my rclone http backend improve that a lot, but still files are not opening
when you posted, you should have been ask for some information.....
I think timeout is more for an idle connection.
I believe you want to increase connection timeouts:
--contimeout duration Connect timeout (default 1m0s)
Does it happen once a minute?
That seems to have fixed the issue for now. I set it for 3 minutes both on the http backend and on the mount.
But I still think it's very odd that changing from an domain to IP had a big change. @ncw does rclone respect the TTL of the domains ?
It's from my understanding that after the domain is resolved to the IP once, it should stay in the cache as long as the TTL allows, so changing from domain to IP shouldn't make any difference
Both the mount and the backend are on the same datacenter too.
As far as I know... What is the ttl of the domain?
When I did that tests it was 5min.
This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.