Date: Thu, 30 Dec 2021 02:13:39 UTC
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=260764 --- Comment #1 from Rick Macklem <rmacklem@FreeBSD.org> --- Well, I think this is easier said than done. Without actually doing testing, I think the timeout is "how long it takes for connect(2) to return" and I think this is the TCP socket option TCP_KEEPINIT. TCP_KEEPINIT can be changed by a setsockopt(), but this is all buried inside the RPC functions in libc and there is no argument to the function to adjust the timeout. --> I think it would involve modifying the RPC library semantics, via some new function that would set the timeout. I will take a closer look, but I'll admit that adding a change like this to the RPC library does not excite me. Would a new NFS mount option that says "do background right away and don't wait for a failure" be sufficient for your case? -- You are receiving this mail because: You are the assignee for the bug.