LU-14834: list the UUIDs of stale clients during recovery

Add a new entry in debugfs for MDT and OSD to display the uuid of clients
yet to be reconnected during a recovery.

For example:

$ lctl get_param obdfilter.lustre-OST0000.recovery_stale_clients
obdfilter.lustre-OST0000.stale_clients=
9a7ab21d-207c-4680-b9bf-b5873fd05540

This will display, during the recovery, the UUIDs of clients
that are expected to be connected.

Signed-off-by: Courrier Guillaume <guillaume.courrier@cea.fr>
Change-Id: Ib8c0b500adc9098e3cfb1998df06757a7d31b7b9
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/44196
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Etienne AUJAMES <eaujames@ddn.com>
Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>

diff --git a/lustre/include/lprocfs_status.h b/lustre/include/lprocfs_status.h
index 6ea80ee..e62cfd1 100644 (file)
--- a/lustre/include/lprocfs_status.h
+++ b/lustre/include/lprocfs_status.h
@@ -678,6 +678,9 @@ void lprocfs_stats_collect(struct lprocfs_stats *stats, int idx,
 /* lprocfs_status.c: recovery status */
 int lprocfs_recovery_status_seq_show(struct seq_file *m, void *data);
 
+/* lprocfs: display the uuid of stale clients */
+int lprocfs_recovery_stale_clients_seq_show(struct seq_file *m, void *data);
+
 /* lprocfs_status.c: hash statistics */
 int lprocfs_hash_seq_show(struct seq_file *m, void *data);
 

diff --git a/lustre/mdt/mdt_lproc.c b/lustre/mdt/mdt_lproc.c
index ccc17a2..b5634ef 100644 (file)
--- a/lustre/mdt/mdt_lproc.c
+++ b/lustre/mdt/mdt_lproc.c
@@ -1605,6 +1605,14 @@ static struct lprocfs_vars lprocfs_mdt_obd_vars[] = {
        { NULL }
 };
 
+LDEBUGFS_SEQ_FOPS_RO_TYPE(mdt, recovery_stale_clients);
+
+static struct ldebugfs_vars ldebugfs_mdt_obd_vars[] = {
+       { .name =       "recovery_stale_clients",
+         .fops =       &mdt_recovery_stale_clients_fops        },
+       { NULL }
+};
+
 static int
 lprocfs_mdt_print_open_files(struct obd_export *exp, void *v)
 {
@@ -1734,6 +1742,7 @@ int mdt_tunables_init(struct mdt_device *mdt, const char *name)
                       mdt_obd_name(mdt), rc);
                return rc;
        }
+       ldebugfs_add_vars(obd->obd_debugfs_entry, ldebugfs_mdt_obd_vars, obd);
 
        rc = tgt_tunables_init(&mdt->mdt_lut);
        if (rc) {

diff --git a/lustre/obdclass/lprocfs_status_server.c b/lustre/obdclass/lprocfs_status_server.c
index 02c3b85..13cd886 100644 (file)
--- a/lustre/obdclass/lprocfs_status_server.c
+++ b/lustre/obdclass/lprocfs_status_server.c
@@ -79,6 +79,42 @@ no_entry:
 }
 EXPORT_SYMBOL(ldebugfs_add_symlink);
 
+int lprocfs_recovery_stale_clients_seq_show(struct seq_file *m, void *data)
+{
+       struct obd_device *obd = m->private;
+       struct obd_export *exp, *n;
+       int connected;
+
+       if (!obd->obd_recovering ||
+           atomic_read(&obd->obd_connected_clients) >=
+           atomic_read(&obd->obd_max_recoverable_clients))
+               /* not in recovery */
+               return 0;
+
+       spin_lock(&obd->obd_dev_lock);
+       list_for_each_entry_safe(exp, n, &obd->obd_exports, exp_obd_chain) {
+               /* don't count self-export as client */
+               if (obd_uuid_equals(&exp->exp_client_uuid,
+                                   &exp->exp_obd->obd_uuid))
+                       continue;
+
+               /* don't count clients which have no slot in last_rcvd
+                * (e.g. lightweight connection)
+                */
+               if (exp->exp_target_data.ted_lr_idx == -1)
+                       continue;
+
+               connected = !exp->exp_failed && (exp->exp_conn_cnt > 0);
+
+               if (!connected)
+                       seq_printf(m, "%s\n", exp->exp_client_uuid.uuid);
+       }
+       spin_unlock(&obd->obd_dev_lock);
+
+       return 0;
+}
+EXPORT_SYMBOL(lprocfs_recovery_stale_clients_seq_show);
+
 #ifdef CONFIG_PROC_FS
 
 int lprocfs_evict_client_open(struct inode *inode, struct file *f)

diff --git a/lustre/ofd/lproc_ofd.c b/lustre/ofd/lproc_ofd.c
index 74020a6..8ddb93f 100644 (file)
--- a/lustre/ofd/lproc_ofd.c
+++ b/lustre/ofd/lproc_ofd.c
@@ -934,6 +934,14 @@ struct lprocfs_vars lprocfs_ofd_obd_vars[] = {
        { NULL }
 };
 
+LDEBUGFS_SEQ_FOPS_RO_TYPE(ofd, recovery_stale_clients);
+
+struct ldebugfs_vars ldebugfs_ofd_obd_vars[] = {
+       { .name =       "recovery_stale_clients",
+         .fops =       &ofd_recovery_stale_clients_fops},
+       { NULL }
+};
+
 /**
  * Initialize OFD statistics counters
  *
@@ -1040,6 +1048,7 @@ int ofd_tunables_init(struct ofd_device *ofd)
                       obd->obd_name, rc);
                RETURN(rc);
        }
+       ldebugfs_add_vars(obd->obd_debugfs_entry, ldebugfs_ofd_obd_vars, obd);
 
        rc = tgt_tunables_init(&ofd->ofd_lut);
        if (rc) {