Performance issues with TRAMP when accessing remote files via SSH

[StefanHeber]

Description:

I really like the Emacs configuration provided in this repository - it’s been incredibly helpful! However, I’ve been experiencing significant performance issues when accessing files on a remote server using TRAMP (e.g., /ssh:user@host:/path/to/file). The process becomes extremely slow, almost to the point of being unusable.

Has anyone else encountered this issue? Are there any recommended solutions or optimizations to improve TRAMP performance in this configuration?

Steps to Reproduce:

Open Emacs with the provided configuration.

Attempt to access a file on a remote server using TRAMP (e.g., C-x C-f /ssh:user@host:/path/to/file RET).

Observe the slow performance when interacting with the remote file.

Expected Behavior:

TRAMP should provide reasonable performance when accessing remote files, allowing for smooth editing and navigation.

Actual Behavior:

TRAMP becomes extremely slow, making it difficult to work with remote files effectively.

Additional Context:

I’m using the latest version of the configuration from this repository.

My Emacs version is 29.3.

The remote server is accessible via SSH without issues outside of Emacs.

Any advice or suggestions would be greatly appreciated! Thank you in advance for your help.

[purcell]

I only occasionally use TRAMP over ssh myself, and I also find it slow, but haven't looked into why. I think that there's just a fair amount of overhead with TRAMP, and its speed may depend on binaries available on the other side, e.g. perl. But I haven't really dug into it. (I tend to use scp: rather than ssh:.)

[dcorking]

@StefanHeber How much faster is Tramp with emacs -Q versus using purcell/emacs.d ?

[StefanHeber]

@StefanHeber How much faster is Tramp with emacs -Q versus using purcell/emacs.d ?

@dcorking Haven't measured it directly, but the difference is quite noticeable. It's worse when accessing something outside the LAN, but even on the local network, there's a clear slowdown.

[purcell]

Then the difference is likely some other stuff running, like flymake, reformatting, git status tracking (diff-hl?) or auto-revert-mode. You might try toggling off some minor modes in an affected buffer and reporting back if that dramatically improves the editing response.