mds: avoid ~mdsdir's scrubbing and reporting damage health status

Setting backtrace.passed as true in case of mdsdir,
as we are not having any backtrace.It would prevent the
entry in the damage table and hence prevent the HEALTH_ERR.

Fixes: https://tracker.ceph.com/issues/58030
Signed-off-by: Neeraj Pratap Singh <neesingh@redhat.com>

diff --git a/src/mds/CInode.cc b/src/mds/CInode.cc
index 4a429f15f2e84acc9d505f5336b155a586e870e7..5fc34a26151400c0f85bad2ab6e24a110a21b459 100644 (file)
--- a/src/mds/CInode.cc
+++ b/src/mds/CInode.cc
@@ -4717,6 +4717,10 @@ void CInode::validate_disk_state(CInode::validated_data *results,
         results->backtrace.error_str << "failed to read off disk; see retval";
         // we probably have a new unwritten file!
         // so skip the backtrace scrub for this entry and say that all's well
+        if (in->is_mdsdir()){
+          dout(20) << "forcing backtrace as passed since mdsdir actually doesn't have backtrace" << dendl;
+          results->backtrace.passed = true;
+        }
         if (in->is_dirty_parent()) {
           dout(20) << "forcing backtrace as passed since inode is dirty parent" << dendl;
           results->backtrace.passed = true;