commit 3584bce5a45f0c40b43385324bfc4a44f1960567
author Amin Hassani <ahassani@google.com> Thu Jun 29 14:21:19 2017 -0700
committer chrome-bot <chrome-bot@chromium.org> Fri Jul 07 15:04:16 2017 -0700
tree 7df432b41d049f943616e38eb4ff503a9412d066
parent 5c02c1335b98b3a1c8759a664bf7b6a48773f913

Retries getting status of update_engine on error before giving up.

Since the update_engine is single threaded, sometimes the operations (like
system write) take too long time and this prevents the update_engine from
resonding to dbus signals. Retrying when the status is "dbus signal failure"
might alleviate the problem. (This CL does not handle the error codes).

BUG=chromium:738024
TEST=none

Change-Id: Ic1b7816b797d84fc6a8c7c7540ddaa22cd3d4d1b
Reviewed-on: https://chromium-review.googlesource.com/556356
Commit-Ready: Grant Grundler <grundler@chromium.org>
Tested-by: Lann Martin <lannm@chromium.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Grant Grundler <grundler@chromium.org>

update_engine_client.cc

diff --git a/update_engine_client.cc b/update_engine_client.cc
index 44897e0..f337dd6 100644
--- a/update_engine_client.cc
+++ b/update_engine_client.cc
@@ -14,6 +14,13 @@
 // limitations under the License.
 //
 
+#include <base/bind.h>
+#include <base/command_line.h>
+#include <base/logging.h>
+#include <base/macros.h>
+#include <brillo/daemons/daemon.h>
+#include <brillo/flag_helper.h>
+
 #include <inttypes.h>
 #include <sysexits.h>
 #include <unistd.h>
@@ -22,13 +29,6 @@
 #include <string>
 #include <vector>
 
-#include <base/bind.h>
-#include <base/command_line.h>
-#include <base/logging.h>
-#include <base/macros.h>
-#include <brillo/daemons/daemon.h>
-#include <brillo/flag_helper.h>
-
 #include "update_engine/client.h"
 #include "update_engine/common/error_code.h"
 #include "update_engine/common/error_code_utils.h"
@@ -52,6 +52,11 @@
 // initialization.
 const int kContinueRunning = -1;
 
+// The ShowStatus request will be retried `kShowStatusRetryCount` times at
+// `kShowStatusRetryInterval` second intervals on failure.
+const int kShowStatusRetryCount = 30;
+const int kShowStatusRetryInterval = 2;
+
 class UpdateEngineClient : public brillo::Daemon {
  public:
   UpdateEngineClient(int argc, char** argv) : argc_(argc), argv_(argv) {
@@ -151,9 +156,17 @@
   string new_version;
   int64_t new_size = 0;
 
-  if (!client_->GetStatus(&last_checked_time, &progress, &current_op,
-                          &new_version, &new_size)) {
-    return false;
+  int retry_count = kShowStatusRetryCount;
+  while (retry_count > 0) {
+    if (client_->GetStatus(&last_checked_time, &progress, &current_op,
+                           &new_version, &new_size)) {
+      break;
+    }
+    if (--retry_count == 0) {
+      return false;
+    }
+    LOG(WARNING) << "Trying " << retry_count << " more times";
+    sleep(kShowStatusRetryInterval);
   }
 
   printf("LAST_CHECKED_TIME=%" PRIi64