rgw/multisite: do not log EBUSY/EAGAIN in per-object sync error log

The bucket-instance level sync (RGWDataSyncSingleEntryCR) excludes EBUSY and
EAGAIN from the sync error log as these are meant to be transient errors that
resolve on retry without an admin intervention.

Also, add EBUSY and EAGAIN to ignore_sync_error(), which is used by the
per-object level sync in RGWSyncObjectCR, to match the bucket-instance
level behavior.

Fixes: https://tracker.ceph.com/issues/76950
Signed-off-by: Oguzhan Ozmen <oozmen@bloomberg.net>

diff --git a/src/rgw/driver/rados/rgw_data_sync.cc b/src/rgw/driver/rados/rgw_data_sync.cc
index 3b742b8570616e6b97c3a5e34b1bd4d02d8f9a12..3901e8c8efc10e2ed0cb6a2c44d064d37ef7ef07 100644 (file)
--- a/src/rgw/driver/rados/rgw_data_sync.cc
+++ b/src/rgw/driver/rados/rgw_data_sync.cc
@@ -4327,6 +4327,8 @@ static bool ignore_sync_error(int err) {
   switch (err) {
     case -ENOENT:
     case -EPERM:
+    case -EBUSY:
+    case -EAGAIN:
       return true;
     default:
       break;