IPVS: one-packet scheduling

Allow one-packet scheduling for UDP connections. When the fwmark-based or normal virtual service is marked with '-o' or '--ops' options all connections are created only to schedule one packet. Useful to schedule UDP packets from same client port to different real servers. Recommended with RR or WRR schedulers (the connections are not visible with ipvsadm -L). Signed-off-by: Nick Chalk <nick@loadbalancer.org> Signed-off-by: Simon Horman <horms@verge.net.au> Signed-off-by: Patrick McHardy <kaber@trash.net>

IPVS: one-packet scheduling
Allow one-packet scheduling for UDP connections. When the fwmark-based or normal virtual service is marked with '-o' or '--ops' options all connections are created only to schedule one packet. Useful to schedule UDP packets from same client port to different real servers. Recommended with RR or WRR schedulers (the connections are not visible with ipvsadm -L). Signed-off-by: Nick Chalk <nick@loadbalancer.org> Signed-off-by: Simon Horman <horms@verge.net.au> Signed-off-by: Patrick McHardy <kaber@trash.net>
Nick Chalk · Patrick McHardy
1 parent c68cd6cc21
Showing 4 changed files with 31 additions and 11 deletions Side-by-side Diff
include/linux/ip_vs.h
net/netfilter/ipvs/ip_vs_conn.c
net/netfilter/ipvs/ip_vs_core.c
net/netfilter/ipvs/ip_vs_ctl.c
@@ -19,6 +19,7 @@
  */
 #define IP_VS_SVC_F_PERSISTENT	0x0001		/* persistent port */
 #define IP_VS_SVC_F_HASHED	0x0002		/* hashed entry */
+#define IP_VS_SVC_F_ONEPACKET	0x0004		/* one-packet scheduling */
  
 /*
  *      Destination Server Flags
@@ -85,6 +86,7 @@
 #define IP_VS_CONN_F_SEQ_MASK	0x0600		/* in/out sequence mask */
 #define IP_VS_CONN_F_NO_CPORT	0x0800		/* no client port set yet */
 #define IP_VS_CONN_F_TEMPLATE	0x1000		/* template, not connection */
+#define IP_VS_CONN_F_ONE_PACKET	0x2000		/* forward only one packet */
  
 #define IP_VS_SCHEDNAME_MAXLEN	16
 #define IP_VS_IFNAME_MAXLEN	16
@@ -158,6 +158,9 @@
 	unsigned hash;
 	int ret;
  
+	if (cp->flags & IP_VS_CONN_F_ONE_PACKET)
+		return 0;
+
 	/* Hash by protocol, client address and port */
 	hash = ip_vs_conn_hashkey(cp->af, cp->protocol, &cp->caddr, cp->cport);
  
@@ -355,8 +358,9 @@
  */
 void ip_vs_conn_put(struct ip_vs_conn *cp)
 {
-	/* reset it expire in its timeout */
-	mod_timer(&cp->timer, jiffies+cp->timeout);
+	unsigned long t = (cp->flags & IP_VS_CONN_F_ONE_PACKET) ?
+		0 : cp->timeout;
+	mod_timer(&cp->timer, jiffies+t);
  
 	__ip_vs_conn_put(cp);
 }
@@ -649,7 +653,7 @@
 	/*
 	 *	unhash it if it is hashed in the conn table
 	 */
-	if (!ip_vs_conn_unhash(cp))
+	if (!ip_vs_conn_unhash(cp) && !(cp->flags & IP_VS_CONN_F_ONE_PACKET))
 		goto expire_later;
  
 	/*
@@ -194,6 +194,7 @@
 	struct ip_vs_dest *dest;
 	struct ip_vs_conn *ct;
 	__be16  dport;			/* destination port to forward */
+	__be16  flags;
 	union nf_inet_addr snet;	/* source network of the client,
 					   after masking */
  
@@ -340,6 +341,10 @@
 		dport = ports[1];
 	}
  
+	flags = (svc->flags & IP_VS_SVC_F_ONEPACKET
+		 && iph.protocol == IPPROTO_UDP)?
+		IP_VS_CONN_F_ONE_PACKET : 0;
+
 	/*
 	 *    Create a new connection according to the template
 	 */
@@ -347,7 +352,7 @@
 			    &iph.saddr, ports[0],
 			    &iph.daddr, ports[1],
 			    &dest->addr, dport,
-			    0,
+			    flags,
 			    dest);
 	if (cp == NULL) {
 		ip_vs_conn_put(ct);
@@ -377,7 +382,7 @@
 	struct ip_vs_conn *cp = NULL;
 	struct ip_vs_iphdr iph;
 	struct ip_vs_dest *dest;
-	__be16 _ports[2], *pptr;
+	__be16 _ports[2], *pptr, flags;
  
 	ip_vs_fill_iphdr(svc->af, skb_network_header(skb), &iph);
 	pptr = skb_header_pointer(skb, iph.len, sizeof(_ports), _ports);
@@ -407,6 +412,10 @@
 		return NULL;
 	}
  
+	flags = (svc->flags & IP_VS_SVC_F_ONEPACKET
+		 && iph.protocol == IPPROTO_UDP)?
+		IP_VS_CONN_F_ONE_PACKET : 0;
+
 	/*
 	 *    Create a connection entry.
 	 */
@@ -414,7 +423,7 @@
 			    &iph.saddr, pptr[0],
 			    &iph.daddr, pptr[1],
 			    &dest->addr, dest->port ? dest->port : pptr[1],
-			    0,
+			    flags,
 			    dest);
 	if (cp == NULL)
 		return NULL;
@@ -464,6 +473,9 @@
 	if (sysctl_ip_vs_cache_bypass && svc->fwmark && unicast) {
 		int ret, cs;
 		struct ip_vs_conn *cp;
+		__u16 flags = (svc->flags & IP_VS_SVC_F_ONEPACKET &&
+				iph.protocol == IPPROTO_UDP)?
+				IP_VS_CONN_F_ONE_PACKET : 0;
 		union nf_inet_addr daddr =  { .all = { 0, 0, 0, 0 } };
  
 		ip_vs_service_put(svc);
@@ -474,7 +486,7 @@
 				    &iph.saddr, pptr[0],
 				    &iph.daddr, pptr[1],
 				    &daddr, 0,
-				    IP_VS_CONN_F_BYPASS,
+				    IP_VS_CONN_F_BYPASS | flags,
 				    NULL);
 		if (cp == NULL)
 			return NF_DROP;
@@ -1864,14 +1864,16 @@
 					   svc->scheduler->name);
 			else
 #endif
-				seq_printf(seq, "%s  %08X:%04X %s ",
+				seq_printf(seq, "%s  %08X:%04X %s %s ",
 					   ip_vs_proto_name(svc->protocol),
 					   ntohl(svc->addr.ip),
 					   ntohs(svc->port),
-					   svc->scheduler->name);
+					   svc->scheduler->name,
+					   (svc->flags & IP_VS_SVC_F_ONEPACKET)?"ops ":"");
 		} else {
-			seq_printf(seq, "FWM  %08X %s ",
-				   svc->fwmark, svc->scheduler->name);
+			seq_printf(seq, "FWM  %08X %s %s",
+				   svc->fwmark, svc->scheduler->name,
+				   (svc->flags & IP_VS_SVC_F_ONEPACKET)?"ops ":"");
 		}
  
 		if (svc->flags & IP_VS_SVC_F_PERSISTENT)
...	...	@@ -19,6 +19,7 @@
19	19	*/
20	20	#define IP_VS_SVC_F_PERSISTENT 0x0001 /* persistent port */
21	21	#define IP_VS_SVC_F_HASHED 0x0002 /* hashed entry */
	22	+#define IP_VS_SVC_F_ONEPACKET 0x0004 /* one-packet scheduling */
22	23
23	24	/*
24	25	* Destination Server Flags
...	...	@@ -85,6 +86,7 @@
85	86	#define IP_VS_CONN_F_SEQ_MASK 0x0600 /* in/out sequence mask */
86	87	#define IP_VS_CONN_F_NO_CPORT 0x0800 /* no client port set yet */
87	88	#define IP_VS_CONN_F_TEMPLATE 0x1000 /* template, not connection */
	89	+#define IP_VS_CONN_F_ONE_PACKET 0x2000 /* forward only one packet */
88	90
89	91	#define IP_VS_SCHEDNAME_MAXLEN 16
90	92	#define IP_VS_IFNAME_MAXLEN 16
...	...	@@ -158,6 +158,9 @@
158	158	unsigned hash;
159	159	int ret;
160	160
	161	+ if (cp->flags & IP_VS_CONN_F_ONE_PACKET)
	162	+ return 0;
	163	+
161	164	/* Hash by protocol, client address and port */
162	165	hash = ip_vs_conn_hashkey(cp->af, cp->protocol, &cp->caddr, cp->cport);
163	166
...	...	@@ -355,8 +358,9 @@
355	358	*/
356	359	void ip_vs_conn_put(struct ip_vs_conn *cp)
357	360	{
358		- /* reset it expire in its timeout */
359		- mod_timer(&cp->timer, jiffies+cp->timeout);
	361	+ unsigned long t = (cp->flags & IP_VS_CONN_F_ONE_PACKET) ?
	362	+ 0 : cp->timeout;
	363	+ mod_timer(&cp->timer, jiffies+t);
360	364
361	365	__ip_vs_conn_put(cp);
362	366	}
...	...	@@ -649,7 +653,7 @@
649	653	/*
650	654	* unhash it if it is hashed in the conn table
651	655	*/
652		- if (!ip_vs_conn_unhash(cp))
	656	+ if (!ip_vs_conn_unhash(cp) && !(cp->flags & IP_VS_CONN_F_ONE_PACKET))
653	657	goto expire_later;
654	658
655	659	/*
...	...	@@ -194,6 +194,7 @@
194	194	struct ip_vs_dest *dest;
195	195	struct ip_vs_conn *ct;
196	196	__be16 dport; /* destination port to forward */
	197	+ __be16 flags;
197	198	union nf_inet_addr snet; /* source network of the client,
198	199	after masking */
199	200
...	...	@@ -340,6 +341,10 @@
340	341	dport = ports[1];
341	342	}
342	343
	344	+ flags = (svc->flags & IP_VS_SVC_F_ONEPACKET
	345	+ && iph.protocol == IPPROTO_UDP)?
	346	+ IP_VS_CONN_F_ONE_PACKET : 0;
	347	+
343	348	/*
344	349	* Create a new connection according to the template
345	350	*/
...	...	@@ -347,7 +352,7 @@
347	352	&iph.saddr, ports[0],
348	353	&iph.daddr, ports[1],
349	354	&dest->addr, dport,
350		- 0,
	355	+ flags,
351	356	dest);
352	357	if (cp == NULL) {
353	358	ip_vs_conn_put(ct);
...	...	@@ -377,7 +382,7 @@
377	382	struct ip_vs_conn *cp = NULL;
378	383	struct ip_vs_iphdr iph;
379	384	struct ip_vs_dest *dest;
380		- __be16 _ports[2], *pptr;
	385	+ __be16 _ports[2], *pptr, flags;
381	386
382	387	ip_vs_fill_iphdr(svc->af, skb_network_header(skb), &iph);
383	388	pptr = skb_header_pointer(skb, iph.len, sizeof(_ports), _ports);
...	...	@@ -407,6 +412,10 @@
407	412	return NULL;
408	413	}
409	414
	415	+ flags = (svc->flags & IP_VS_SVC_F_ONEPACKET
	416	+ && iph.protocol == IPPROTO_UDP)?
	417	+ IP_VS_CONN_F_ONE_PACKET : 0;
	418	+
410	419	/*
411	420	* Create a connection entry.
412	421	*/
...	...	@@ -414,7 +423,7 @@
414	423	&iph.saddr, pptr[0],
415	424	&iph.daddr, pptr[1],
416	425	&dest->addr, dest->port ? dest->port : pptr[1],
417		- 0,
	426	+ flags,
418	427	dest);
419	428	if (cp == NULL)
420	429	return NULL;
...	...	@@ -464,6 +473,9 @@
464	473	if (sysctl_ip_vs_cache_bypass && svc->fwmark && unicast) {
465	474	int ret, cs;
466	475	struct ip_vs_conn *cp;
	476	+ __u16 flags = (svc->flags & IP_VS_SVC_F_ONEPACKET &&
	477	+ iph.protocol == IPPROTO_UDP)?
	478	+ IP_VS_CONN_F_ONE_PACKET : 0;
467	479	union nf_inet_addr daddr = { .all = { 0, 0, 0, 0 } };
468	480
469	481	ip_vs_service_put(svc);
...	...	@@ -474,7 +486,7 @@
474	486	&iph.saddr, pptr[0],
475	487	&iph.daddr, pptr[1],
476	488	&daddr, 0,
477		- IP_VS_CONN_F_BYPASS,
	489	+ IP_VS_CONN_F_BYPASS \| flags,
478	490	NULL);
479	491	if (cp == NULL)
480	492	return NF_DROP;
...	...	@@ -1864,14 +1864,16 @@
1864	1864	svc->scheduler->name);
1865	1865	else
1866	1866	#endif
1867		- seq_printf(seq, "%s %08X:%04X %s ",
	1867	+ seq_printf(seq, "%s %08X:%04X %s %s ",
1868	1868	ip_vs_proto_name(svc->protocol),
1869	1869	ntohl(svc->addr.ip),
1870	1870	ntohs(svc->port),
1871		- svc->scheduler->name);
	1871	+ svc->scheduler->name,
	1872	+ (svc->flags & IP_VS_SVC_F_ONEPACKET)?"ops ":"");
1872	1873	} else {
1873		- seq_printf(seq, "FWM %08X %s ",
1874		- svc->fwmark, svc->scheduler->name);
	1874	+ seq_printf(seq, "FWM %08X %s %s",
	1875	+ svc->fwmark, svc->scheduler->name,
	1876	+ (svc->flags & IP_VS_SVC_F_ONEPACKET)?"ops ":"");
1875	1877	}
1876	1878
1877	1879	if (svc->flags & IP_VS_SVC_F_PERSISTENT)