If any backup process failed, try again.

Try again without waiting for an update of the fetched file, which
won't happen in the tests (but probably will happen in real life,
masking this bug).

1 files changed, 7 insertions, 3 deletions

diff --git a/tools/merge_backup.py b/tools/merge_backup.py
index 56b5b57..12ff99d 100755
--- a/tools/merge_backup.py
+++ b/tools/merge_backup.py
@@ -155,6 +155,7 @@ def update_backupfile(mergedb, nodename, tree_size, root_hash):
     write_file(backuppath, backupdata)
 
 def merge_backup(args, config, localconfig, secondaries):
+    failures = 0
     paths = localconfig["paths"]
     mergedb = paths["mergedb"]
     chainsdb = perm(localconfig.get("dbbackend", "filedb"), mergedb + "/chains")
@@ -200,6 +201,7 @@ def merge_backup(args, config, localconfig, secondaries):
                         update_backupfile(mergedb, nodename, tree_size, root_hash)
                     else:
                         logging.warning("%s failure: %d", nodename, p.exitcode)
+                        failures += 1
                     del procs[p]
             if not procs:
                 break
@@ -207,7 +209,7 @@ def merge_backup(args, config, localconfig, secondaries):
 
     if args.timing:
         logging.debug("timing: merge_backup: %s", timing["deltatimes"])
-    return 0
+    return failures
 
 def main():
     """
@@ -247,12 +249,14 @@ def main():
     fetched_statinfo = waitforfile(fetched_path)
     retval = 0
     while True:
-        retval = merge_backup(args, config, localconfig, nodes)
-        if retval or not args.mergeinterval:
+        failures = merge_backup(args, config, localconfig, nodes)
+        if not args.mergeinterval:
             break
         fetched_statinfo_old = fetched_statinfo
         while fetched_statinfo == fetched_statinfo_old:
             sleep(max(3, args.mergeinterval / 10))
+            if failures > 0:
+                break
             fetched_statinfo = stat(fetched_path)
 
     return retval