You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We've started to use the RISCV runners (thank you!) and one thing we're seeing is that on several common bandwidth-bound workloads (GHA cache and GHCR images) we're seeing very slow network performance -- on the order of 3MB/s. This makes lots of common workflows incredibly slow.
Is there anything that can be done to improve thsi?
Hi @alex , thanks for using this service!
The machines we currently have access to have 100Mbs network connections. Also, they are not hosted by GitHub so don't have closer network access. However, 3MB/s seems on the slower side of things, it's usually closer to 7-8MB/s. Would you have a link to such workflow? I can have a look at whether something is wrong with the underlying node.
Activity
luhenry commentedon Oct 28, 2026
Hi @alex , thanks for using this service!
The machines we currently have access to have 100Mbs network connections. Also, they are not hosted by GitHub so don't have closer network access. However, 3MB/s seems on the slower side of things, it's usually closer to 7-8MB/s. Would you have a link to such workflow? I can have a look at whether something is wrong with the underlying node.
alex commentedon Oct 29, 2026
Sure, happy to point at jobs -- it's easiest to see the network speed on the rust-cache stage: