Skip to contentSkip to conversation

xtables: Add locking to prevent concurrent instances

PATCH Superseded Phil Oester 2 participants 4 messages May 27, 2013 – May 29, 2013 netfilter

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

PATCH v3 1/1 xtables: Add locking to prevent concurrent instances

There have been numerous complaints and bug reports over the years when admins attempt to run more than one instance of iptables simultaneously. Currently open bug reports which are related:

325: Parallel execution of the iptables is impossible 758: Retry iptables command on transient failure 764: Doing -Z twice in parallel breaks counters 822: iptables shows negative or other bad packet/byte counts

As Patrick notes in 325: "Since this has been a problem people keep running into, I'd suggest to simply add some locking to iptables to catch the most common case."

I started looking into alternatives to add locking, and of course the most common/obvious solution is to use a pidfile. But this has various downsides, such as if the application is terminated abnormally and the pidfile isn't cleaned up. And this also requires a writable filesystem. Using a UNIX domain socket file (e.g. in /var/run) has similar issues.

Starting in 2.2, Linux added support for abstract sockets. These sockets require no filesystem, and automatically disappear once the application terminates. This is the locking solution I chose to implement in xtables-multi. As an added bonus, since each network namespace has its own socket pool, an iptables instance running in one namespace will not lock out an iptables instance running in another namespace. A filesystem approach would have to recognize and handle multiple network namespaces.

Phil

---

v2: Addressed Patrick's comments - locking attempts will be made indefinitely until successful
v3: Update warning message to more closely resemble locking output from yum

iptables/xtables-multi.c +29 -0
index 8014d5f..56d2c91 100644
@@ -1,8 +1,12 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <sys/socket.h>
+#include <sys/un.h>
+#include <unistd.h>
 #include "xshared.h"
 
+#include "xtables.h"
 #include "xtables-multi.h"
 
 #ifdef ENABLE_IPV4
@@ -35,7 +39,32 @@ static const struct subcommand multi_subcommands[] = {
 	{NULL},
 };
 
+#define XTMSOCKET_NAME "xtables_multi"
+#define XTMSOCKET_LEN 14
+
 int main(int argc, char **argv)
 {
+	int i = 0, ret, xtm_socket;
+	struct sockaddr_un xtm_addr;
+
+	memset(&xtm_addr, 0, sizeof(xtm_addr));
+	xtm_addr.sun_family = AF_UNIX;
+	strcpy(xtm_addr.sun_path+1, XTMSOCKET_NAME);
+	xtm_socket = socket(AF_UNIX, SOCK_STREAM, 0);
+	/* If we can't even create a socket, just revert to prior (lockless) behavior */
+	if (xtm_socket < 0)
+		return subcmd_main(argc, argv, multi_subcommands);
+
+	while (1) {
+		ret = bind(xtm_socket, (struct sockaddr*)&xtm_addr,
+			   offsetof(struct sockaddr_un, sun_path)+XTMSOCKET_LEN);
+		if (ret == 0)
+			break;
+		if (++i % 5 == 0)
+			fprintf(stderr, "Another app is currently holding the xtables lock; "
+				"waiting for it to exit...\n");
+		sleep(1);
+	}
+
 	return subcmd_main(argc, argv, multi_subcommands);
 }
#1 lore raw

Hi Phil,

Thanks for looking at this old issue and bring it back to discussion.

Quoted 75 lines · Phil Oester

There have been numerous complaints and bug reports over the years when admins attempt to run more than one instance of iptables simultaneously. Currently open bug reports which are related:

325: Parallel execution of the iptables is impossible 758: Retry iptables command on transient failure 764: Doing -Z twice in parallel breaks counters 822: iptables shows negative or other bad packet/byte counts

As Patrick notes in 325: "Since this has been a problem people keep running into, I'd suggest to simply add some locking to iptables to catch the most common case."

I started looking into alternatives to add locking, and of course the most common/obvious solution is to use a pidfile. But this has various downsides, such as if the application is terminated abnormally and the pidfile isn't cleaned up. And this also requires a writable filesystem. Using a UNIX domain socket file (e.g. in /var/run) has similar issues.

Starting in 2.2, Linux added support for abstract sockets. These sockets require no filesystem, and automatically disappear once the application terminates. This is the locking solution I chose to implement in xtables-multi. As an added bonus, since each network namespace has its own socket pool, an iptables instance running in one namespace will not lock out an iptables instance running in another namespace. A filesystem approach would have to recognize and handle multiple network namespaces.

Phil

---

v2: Addressed Patrick's comments - locking attempts will be made indefinitely until successful
v3: Update warning message to more closely resemble locking output from yum

iptables/xtables-multi.c +19 -0
index 8014d5f..56d2c91 100644
@@ -1,8 +1,12 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <sys/socket.h>
+#include <sys/un.h>
+#include <unistd.h>
 #include "xshared.h"
 
+#include "xtables.h"
 #include "xtables-multi.h"
 
 #ifdef ENABLE_IPV4
@@ -35,7 +39,32 @@ static const struct subcommand multi_subcommands[] = {
 	{NULL},
 };
 
+#define XTMSOCKET_NAME "xtables_multi"
+#define XTMSOCKET_LEN 14
+
 int main(int argc, char **argv)
 {
+	int i = 0, ret, xtm_socket;
+	struct sockaddr_un xtm_addr;
+
+	memset(&xtm_addr, 0, sizeof(xtm_addr));
+	xtm_addr.sun_family = AF_UNIX;
+	strcpy(xtm_addr.sun_path+1, XTMSOCKET_NAME);
+	xtm_socket = socket(AF_UNIX, SOCK_STREAM, 0);
+	/* If we can't even create a socket, just revert to prior (lockless) behavior */
+	if (xtm_socket < 0)
+		return subcmd_main(argc, argv, multi_subcommands);
+
+	while (1) {

iptables/xtables-multi.c static const struct subcommand multi_subcommands[] = { · line 68

Quoted 10 lines
iptables/xtables-multi.c +10 -0
+		ret = bind(xtm_socket, (struct sockaddr*)&xtm_addr,
+			   offsetof(struct sockaddr_un, sun_path)+XTMSOCKET_LEN);
+		if (ret == 0)
+			break;
+		if (++i % 5 == 0)
+			fprintf(stderr, "Another app is currently holding the xtables lock; "
+				"waiting for it to exit...\n");
+		sleep(1);
+	}
+

I think we can:

* Add a new option to explicitly request this behaviour, just as a way to assert that you really want iptables to retry. Harald was rising some concerns on the expected results in case of clash that sound reasonable to me.

* Limit this to ip[6]tables. All bug reports refer to it.

Regards.

#2 lore raw
Quoted 6 lines · Pablo Neira Ayuso

I think we can:

* Add a new option to explicitly request this behaviour, just as a way to assert that you really want iptables to retry. Harald was rising some concerns on the expected results in case of clash that sound reasonable to me.

I agree that the retry behaviour could be made optional, however I'm not sure that the locking behaviour should be optional. It leads to various races, some of which are subtle and can occur during if-up and other "behind the scenes" scenarios. In my personal experience, I had to implement locking inside my scripts because I was hitting races fairly regularly with dynamic rule additions/deletions (and I suspect other admins have done the same). Perhaps we can change the error to say:

"Another app is currently holding the ip[6]tables lock (use -r option to enable retries)"

or something similar? At least that is more informative than the typical race error of "iptables: Resource temporarily unavailable".

Quoted 1 line

* Limit this to ip[6]tables. All bug reports refer to it.

Seems reasonable. If future races are discovered in -save or -restore, it could be easily changed.

Phil

#3 lore raw
Quoted 17 lines · Phil Oester
Quoted 6 lines · Pablo Neira Ayuso

I think we can:

* Add a new option to explicitly request this behaviour, just as a way to assert that you really want iptables to retry. Harald was rising some concerns on the expected results in case of clash that sound reasonable to me.

I agree that the retry behaviour could be made optional, however I'm not sure that the locking behaviour should be optional. It leads to various races, some of which are subtle and can occur during if-up and other "behind the scenes" scenarios. In my personal experience, I had to implement locking inside my scripts because I was hitting races fairly regularly with dynamic rule additions/deletions (and I suspect other admins have done the same). Perhaps we can change the error to say:

Quoted 1 line

"Another app is currently holding the ip[6]tables lock (use -r option to enable retries)"

That seems reasonable to me.

Quoted 7 lines

or something similar? At least that is more informative than the typical race error of "iptables: Resource temporarily unavailable".

Quoted 1 line

* Limit this to ip[6]tables. All bug reports refer to it.

Seems reasonable. If future races are discovered in -save or -restore, it could be easily changed.

Good. Would you send us a new patch?

Thanks.