Zero downtime restarts for golang HTTP and HTTPS servers.

Well... it's what you want right - no need to hook in and out on a loadbalancer or something - just compile, SIGHUP, start new one, finish old requests etc.

There is https://github.com/rcrowley/goagain and i looked at https://fitstar.github.io/falcore/hot_restart.html which looked easier to do, but still some assembly required. I wanted something that's ideally as simple as

err := endless.ListenAndServe("localhost:4242", mux)

I found the excellent post Graceful Restart in Golang by Grisha Trubetskoy and took his code as a start. So a lot of credit to Grisha!

• Drop-in replacement for http.ListenAndServe and http.ListenAndServeTLS
• Signal hooks to execute your own code before or after the listened to signals (SIGHUP, SIGUSR1, SIGUSR2, SIGINT, SIGTERM, SIGTSTP)
• You can start multiple servers from one binary and endless will take care of the different sockets/ports assignments when restarting

There are three variables exported by the package that control the values set for DefaultReadTimeOut, DefaultWriteTimeOut, and MaxHeaderBytes on the inner http.Server:

DefaultReadTimeOut    time.Duration
DefaultWriteTimeOut   time.Duration
DefaultMaxHeaderBytes int

The endless default behaviour is to use the same defaults defined in net/http.

These have impact on endless by potentially not letting the parent process die until all connections are handled/finished.

To deal with hanging requests on the parent after restarting endless will hammer the parent 60 seconds after receiving the shutdown signal from the forked child process. When hammered still running requests get terminated. This behaviour can be controlled by another exported variable:

DefaultHammerTime time.Duration

The default is 60 seconds. When set to -1 hammerTime() is not invoked automatically. You can then hammer the parent manually by sending SIGUSR2. This will only hammer the parent if it is already in shutdown mode. So unless the process had received a SIGTERM, SIGSTOP, or SIGINT (manually or by forking) before SIGUSR2 will be ignored.

If you had hanging requests and the server got hammered you will see a log message like this:

2015/04/04 13:04:10 [STOP - Hammer Time] Forcefully shutting down parent

import "github.com/fvbock/endless"

and then replacing http.ListenAndServe with endless.ListenAndServe or http.ListenAndServeTLS with endless.ListenAndServeTLS

err := endless.ListenAndServe("localhost:4242", handler)

After starting your server you can make some changes, build, and send SIGHUP to the running process and it will finish handling any outstanding requests and serve all new incoming ones with the new binary.

More examples are in here

There is also GoDoc Documentation

The endless server will listen for the following signals: syscall.SIGHUP, syscall.SIGUSR1, syscall.SIGUSR2, syscall.SIGINT, syscall.SIGTERM, and syscall.SIGTSTP:

SIGHUP will trigger a fork/restart

syscall.SIGINT and syscall.SIGTERM will trigger a shutdown of the server (it will finish running requests)

SIGUSR2 will trigger hammerTime

SIGUSR1 and SIGTSTP are listened for but do not trigger anything in the endless server itself. (probably useless - might get rid of those two)

You can hook your own functions to be called pre or post signal handling - eg. pre fork or pre shutdown. More about that in the hook example.

Currently you cannot restart a server on a different port than the previous version was running on.

If you want to save actual pid file, you can change the BeforeBegin hook like this:

server := endless.NewServer("localhost:4242", handler)
server.BeforeBegin = func(add string) {
	log.Printf("Actual pid is %d", syscall.Getpid())
	// save it somehow
}
err := server.ListenAndServe()

• tests
• documentation
• less ugly wrapping of the tls.listener