#!/bin/bash . $(dirname $0)/../../include.rc . $(dirname $0)/../../volume.rc cleanup; TEST glusterd TEST pidof glusterd TEST $CLI volume create $V0 replica 2 $H0:$B0/${V0}{1,2} EXPECT 'Created' volinfo_field $V0 'Status'; TEST $CLI volume set $V0 performance.stat-prefetch off TEST $CLI volume start $V0 EXPECT 'Started' volinfo_field $V0 'Status'; TEST glusterfs --direct-io-mode=yes --use-readdirp=no --entry-timeout=0 --attribute-timeout=0 -s $H0 --volfile-id $V0 $M0; mkdir -p $M0/dir1/dir2 TEST rm -f $(gf_get_gfid_backend_file_path $B0/${V0}1 "dir1") TEST rmdir $B0/${V0}1/dir1/dir2 TEST stat $M0/dir1/dir2 TEST [ -d $B0/${V0}1/dir1/dir2 ] TEST [ ! -d $(gf_get_gfid_backend_file_path $B0/${V0}1 "dir1") ] # Stop the volume to flush caches and force symlink recreation EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $M0 TEST $CLI volume stop $V0 EXPECT 'Stopped' volinfo_field $V0 'Status'; TEST $CLI volume start $V0 EXPECT 'Started' volinfo_field $V0 'Status'; TEST glusterfs --entry-timeout=0 --attribute-timeout=0 -s $H0 --volfile-id $V0 $M0; # Till now, protocol/server was not doing inode linking as part of readdirp. # But pas part of user servicable snapshots patcth, changes to do inode linking # in protocol/server in readdirp, were introduced. So now to make sure # the gfid handle of dir1 is healed, explicit lookup has to be sent on it. # Otherwise, whenever ls -l is done just on the mount point $M0, lookup on the # entries received as part of readdirp, is not sent, because the inodes for # those entries were linked as part of readdirp itself. i.e instead of doing # "ls -l $M0", it has to be the below command. ls -l $M0/dir1; TEST [ -h $(gf_get_gfid_backend_file_path $B0/${V0}1 "dir1") ] EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $M0 TEST $CLI volume stop $V0 TEST $CLI volume delete $V0 cleanup