Hey Brian,
I just pushed some improvements to the MDS lookuphash that should fix the
NFS issues. The first piece will search for an ino in other MDS's caches
(this actually needed for cross-mds unlink/rename as well). The other
piece looks at the on-disk directory objects to discover the ino path and
then does a traverse. Both methods are now wired up to lookuphash.
Once you have a test case that can reliably reproduce the NFS ESTALE,
let's see if this fixes it! (You may have to disable it with something
like the below to get the test case right. The improvements are now in
the master branch of ceph.git.)
Thanks!
sage
diff --git a/src/mds/Server.cc b/src/mds/Server.cc
index 58be710..0b36011 100644
--- a/src/mds/Server.cc
+++ b/src/mds/Server.cc
@@ -2065,8 +2065,9 @@ void Server::handle_client_lookup_hash(MDRequest
*mdr)
// try the directory
CInode *diri = mdcache->get_inode(req->get_filepath2().get_ino());
if (!diri) {
- mdcache->find_ino_peers(req->get_filepath2().get_ino(),
- new C_MDS_LookupHash(this, mdr), -1);
+ reply_request(mdr, -ESTALE);
+ /* mdcache->find_ino_peers(req->get_filepath2().get_ino(),
+ new C_MDS_LookupHash(this, mdr), -1);*/
return;
}
if (diri->state_test(CInode::STATE_PURGING)) {
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
