Description of problem: I hit this corruption while running revolver on the smoke cluster. It is running the latest bits (kernel and GFS2) from Steve's nmw git tree. GFS2: fsid=Smoke_Cluster:gfs2.1.2: fatal: invalid metadata block GFS2: fsid=Smoke_Cluster:gfs2.1.2: bh = 10774747 (magic number) GFS2: fsid=Smoke_Cluster:gfs2.1.2: function = get_leaf, file = fs/gfs2/dir.c, line = 701 GFS2: fsid=Smoke_Cluster:gfs2.1.2: about to withdraw this file system GFS2: fsid=Smoke_Cluster:gfs2.1.2: telling LM to withdraw [<c0104caf>] show_trace+0x12/0x14 [<c0104d08>] dump_stack+0x16/0x18 [<e0547a09>] gfs2_assert_withdraw_i+0x43/0x50 [gfs2] [<e053680e>] log_bmap+0xa6/0xe6 [gfs2] [<e0537177>] gfs2_log_fake_buf+0x19/0x115 [gfs2] [<e0537b1d>] buf_lo_before_commit+0x12a/0x17e [gfs2] [<e0536f7b>] gfs2_log_flush+0x119/0x2c7 [gfs2] [<e053f541>] gfs2_sync_fs+0x18/0x1c [gfs2] [<c0156475>] __fsync_super+0x42/0x5c [<c01700b3>] freeze_bdev+0x3c/0x6c [<c0381afe>] dm_suspend+0xf1/0x268 [<c0384371>] dev_suspend+0x55/0x163 [<c0384ccf>] ctl_ioctl+0x220/0x261 [<c015e88f>] do_ioctl+0x8b/0xa3 [<c015eaf0>] vfs_ioctl+0x249/0x25c [<c015eb34>] sys_ioctl+0x31/0x49 [<c01032b2>] sysenter_past_esp+0x5f/0x85 ======================= GFS2: fsid=Smoke_Cluster:gfs2.1.2: withdrawn [<c0104260>] show_trace_log_lvl+0x1a/0x2f [<c0104caf>] show_trace+0x12/0x14 [<c0104d08>] dump_stack+0x16/0x18 [<e0535f8a>] gfs2_lm_withdraw+0x81/0x8d [gfs2] [<e0547891>] gfs2_meta_check_ii+0x51/0x5e [gfs2] [<e052b6ee>] get_leaf+0x75/0xbd [gfs2] [<e052bcd8>] get_first_leaf+0x28/0x2f [gfs2] [<e052bd68>] gfs2_dirent_search+0x89/0x1a1 [gfs2] [<e052d45f>] gfs2_dir_search+0x21/0x6b [gfs2] [<e0534629>] gfs2_lookupi+0x11e/0x155 [gfs2] [<e053e7e1>] gfs2_lookup+0x1d/0x4d [gfs2] [<c015b9c1>] lookup_hash+0x83/0x9b [<c015d90c>] open_namei+0xe1/0x558 [<c0153bca>] do_filp_open+0x26/0x3b [<c0153c22>] do_sys_open+0x43/0xc7 [<c0153cde>] sys_open+0x1c/0x1e [<c01032b2>] sysenter_past_esp+0x5f/0x85 =======================
Blocking recovery tests at the moment, needs to be addressed prior to release.
This request was evaluated by Red Hat Product Management for inclusion in a Red Hat Enterprise Linux maintenance release. Product Management has requested further review of this request by Red Hat Engineering, for potential inclusion in a Red Hat Enterprise Linux Update release for currently deployed products. This request is not yet committed for inclusion in an Update release.
Abhi, Can you recreate this and leave it in the failing state for me? I want to see the cluster after a failure so I can examine the journals.
Abhi, can you confirm that you never see this bug before at least one node has completed recovery?
This bug is the same as 248176. *** This bug has been marked as a duplicate of 248176 ***