osd: refuse to start if configure object limits don't work

Fail early if the configured max object name cannot be stored on
the (FileStore) backend.  This avoids rude surprises down the
line when some random rados op get ENAMETOOLONG.

Signed-off-by: Sage Weil <sage@redhat.com>

diff --git a/src/osd/OSD.cc b/src/osd/OSD.cc
index 7c260ff39aeae99b023af98aad36bb1674af586d..b37134114e34e6db45da9239a611c17dfd6d590b 100644 (file)
--- a/src/osd/OSD.cc
+++ b/src/osd/OSD.cc
@@ -2022,6 +2022,25 @@ int OSD::init()
   int rotating_auth_attempts = 0;
   const int max_rotating_auth_attempts = 10;
 
+  // sanity check long object name handling
+  {
+    hobject_t l;
+    l.oid.name = string(g_conf->osd_max_object_name_len, 'n');
+    l.set_key(string(g_conf->osd_max_object_name_len, 'k'));
+    l.nspace = string(g_conf->osd_max_object_namespace_len, 's');
+    r = store->validate_hobject_key(l);
+    if (r < 0) {
+      derr << "backend (" << store->get_type() << ") is unable to support max "
+          << "object name[space] len" << dendl;
+      derr << "   osd max object name len = "
+          << g_conf->osd_max_object_name_len << dendl;
+      derr << "   osd max object namespace len = "
+          << g_conf->osd_max_object_namespace_len << dendl;
+      derr << cpp_strerror(r) << dendl;
+      goto out;
+    }
+  }
+
   // read superblock
   r = read_superblock();
   if (r < 0) {