Adding a callback for incoming sm_notify to better handle stale lock issue in client crash recovery in HA-NFS environment. The callout parameters are:

Commit Message

[e-mail] May 8, 2014, 12:16 a.m. UTC

From: rong zeng <[e-mail]>

1. "sm-notify" - callout name
2. monitored client name as in the SM_NOTIFY request
3. IP of the sender of the SM_NOITFY request.
4. state value in the SM_NOTIFY request

This new interface can be used by different HA-NFS product
in its specific configuration and environment to
recover from the client crash and stale lock scenarios.

Signed-off-by: Rong Zeng <[e-mail]>
---
 utils/statd/callback.c |   13 ++++++++-----
 utils/statd/statd.man  |   20 ++++++++++++++++----
 2 files changed, 24 insertions(+), 9 deletions(-)

Patch

diff --git a/utils/statd/callback.c b/utils/statd/callback.c
index d1cc139..58ba70b 100644
--- a/utils/statd/callback.c
+++ b/utils/statd/callback.c
@@ -15,6 +15,7 @@ 
 #include "rpcmisc.h"
 #include "statd.h"
 #include "notlist.h"
+#include "ha-callout.h"

 /* Callback notify list. */
 /* notify_list *cbnl = NULL; ... never used */
@@ -87,6 +88,13 @@  sm_notify_1_svc(struct stat_chge *argp, struct svc_req *rqstp)
 	xlog(D_CALL, "Received SM_NOTIFY from %s, state: %d",
 				argp->mon_name, argp->state);

+	if (!statd_present_address(sap, ip_addr, sizeof(ip_addr))) {
+		xlog_warn("Unrecognized sender address");
+		return ((void *) &result);
+	}
+
+	ha_callout("sm-notify", argp->mon_name, ip_addr, argp->state);
+
 	/* quick check - don't bother if we're not monitoring anyone */
 	if (rtnl == NULL) {
 		xlog_warn("SM_NOTIFY from %s while not monitoring any hosts",
@@ -94,11 +102,6 @@  sm_notify_1_svc(struct stat_chge *argp, struct svc_req *rqstp)
 		return ((void *) &result);
 	}

-	if (!statd_present_address(sap, ip_addr, sizeof(ip_addr))) {
-		xlog_warn("Unrecognized sender address");
-		return ((void *) &result);
-	}
-
 	/* okir change: statd doesn't remove the remote host from its
 	 * internal monitor list when receiving an SM_NOTIFY call from
 	 * it. Lockd will want to continue monitoring the remote host
diff --git a/utils/statd/statd.man b/utils/statd/statd.man
index 896c2f8..1e5520c 100644
--- a/utils/statd/statd.man
+++ b/utils/statd/statd.man
@@ -346,7 +346,8 @@  points due to inactivity.
 .SS High-availability callouts
 .B rpc.statd
 can exec a special callout program during processing of
-successful SM_MON, SM_UNMON, and SM_UNMON_ALL requests.
+successful SM_MON, SM_UNMON, and SM_UNMON_ALL requests,
+or when it receives SM_NOTIFY.
 Such a program may be used in High Availability NFS (HA-NFS)
 environments to track lock state that may need to be migrated after
 a system reboot.
@@ -357,15 +358,26 @@  option.
 The program is run with 3 arguments:
 The first is either
 .B add-client
-or
 .B del-client
+or
+.B sm-notify
 depending on the reason for the callout.
 The second is the
 .I mon_name
 of the monitored peer.
 The third is the
-.I caller_name
-of the requesting lock manager.
+.I caller_name 
+of the requesting lock manager for
+.B add-client
+or
+.B del-client
+, otherwise it is
+.I IP_address 
+of the caller sending SM_NOTIFY.
+The forth is the
+.I state_value
+in the SM_NOTIFY request.
+
 .SS IPv6 and TI-RPC support
 TI-RPC is a pre-requisite for supporting NFS on IPv6.
 If TI-RPC support is built into