Unless developers double as deployment gnomes or devops devoperators, they have no need to be on the production servers if they are not actively engaged in debugging a production issue that does not manifest on the dev or other) tier.

I personally find that if something goes wrong, the blame gang goes searching for developer that obviously broke it.

I was once publishing new web content via FTP to an IIS server when there was a disruption in mainframe connectivity to our data center. I was asked by more than one manager, in earnest, why and how I broke the 3270 connectivity. Note: I had no mainframe access at all, none, nada, zilch, zip.

IT's much easier to say "I don't have access to that machine, sorry."