Merge pull request #5014 from MichaelEischer/configurable-slow-request-timeout

Make timeout for slow requests configurable

changelog/unreleased/issue-4970

@@ -0,0 +1,13 @@
+Enhancement: Make timeout for stuck requests customizable
+
+Restic monitors connections to the backend to detect stuck requests. If a request
+does not return any data within five minutes, restic assumes the request is stuck and
+retries it. However, for large repositories it sometimes takes longer than that to
+collect a list of all files, causing the following error:
+
+`List(data) returned error, retrying after 1s: [...]: request timeout`
+
+It is now possible to increase the timeout using the `--stuck-request-timeout` option.
+
+https://github.com/restic/restic/issues/4970
+https://github.com/restic/restic/pull/5014

cmd/restic/global.go

@@ -140,6 +140,7 @@ func init() {
 	f.UintVar(&globalOptions.PackSize, "pack-size", 0, "set target pack `size` in MiB, created pack files may be larger (default: $RESTIC_PACK_SIZE)")
 	f.StringSliceVarP(&globalOptions.Options, "option", "o", []string{}, "set extended option (`key=value`, can be specified multiple times)")
 	f.StringVar(&globalOptions.HTTPUserAgent, "http-user-agent", "", "set a http user agent for outgoing http requests")
+	f.DurationVar(&globalOptions.StuckRequestTimeout, "stuck-request-timeout", 5*time.Minute, "`duration` after which to retry stuck requests")
 	// Use our "generate" command instead of the cobra provided "completion" command
 	cmdRoot.CompletionOptions.DisableDefaultCmd = true
 

doc/faq.rst

@@ -228,3 +228,17 @@ Restic backup command fails to find a valid file in Windows
 
 If the name of a file in Windows contains an invalid character, Restic will not be
 able to read the file. To solve this issue, consider renaming the particular file.
+
+What can I do in case of "request timeout" errors?
+--------------------------------------------------
+
+Restic monitors connections to the backend to detect stuck requests. If a request
+does not return any data within five minutes, restic assumes the request is stuck and
+retries it. However, for large repositories it sometimes takes longer than that to
+collect a list of all files, causing the following error:
+
+::
+
+    List(data) returned error, retrying after 1s: [...]: request timeout
+
+In this case you can increase the timeout using the ``--stuck-request-timeout`` option.

internal/backend/http_transport.go

@@ -31,6 +31,9 @@ type TransportOptions struct {
 
 	// Specify Custom User-Agent for the http Client
 	HTTPUserAgent string
+
+	// Timeout after which to retry stuck requests
+	StuckRequestTimeout time.Duration
 }
 
 // readPEMCertKey reads a file and returns the PEM encoded certificate and key
@@ -143,7 +146,11 @@ func Transport(opts TransportOptions) (http.RoundTripper, error) {
 	}
 
 	if feature.Flag.Enabled(feature.BackendErrorRedesign) {
-		rt = newWatchdogRoundtripper(rt, 5*time.Minute, 128*1024)
+		if opts.StuckRequestTimeout == 0 {
+			opts.StuckRequestTimeout = 5 * time.Minute
+		}
+
+		rt = newWatchdogRoundtripper(rt, opts.StuckRequestTimeout, 128*1024)
 	}
 
 	// wrap in the debug round tripper (if active)