[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Xen-changelog] [xen-unstable] Remus: increase failover timeout from 500ms to 1s



# HG changeset patch
# User Keir Fraser <keir.fraser@xxxxxxxxxx>
# Date 1265966590 0
# Node ID a948403c8f99013cd7bfd5c441e84c41a0e4009e
# Parent  a4ddf47a1e912d18f982fded29539a69e9ac1b0c
Remus: increase failover timeout from 500ms to 1s

500ms is aggressive enough to trigger split-brain under fairly
ordinary workloads, particularly for HVM. The long-term fix is to
integrate with a real HA monitor like linux HA.

Signed-off-by: Brendan Cully <brendan@xxxxxxxxx>
---
 tools/blktap2/drivers/block-remus.c |   10 +++++-----
 tools/libxc/xc_domain_restore.c     |    6 +++---
 2 files changed, 8 insertions(+), 8 deletions(-)

diff -r a4ddf47a1e91 -r a948403c8f99 tools/blktap2/drivers/block-remus.c
--- a/tools/blktap2/drivers/block-remus.c       Fri Feb 12 09:21:57 2010 +0000
+++ b/tools/blktap2/drivers/block-remus.c       Fri Feb 12 09:23:10 2010 +0000
@@ -59,7 +59,7 @@
 #include <unistd.h>
 
 /* timeout for reads and writes in ms */
-#define NET_TIMEOUT 500
+#define HEARTBEAT_MS 1000
 #define RAMDISK_HASHSIZE 128
 
 /* connect retry timeout (seconds) */
@@ -604,8 +604,8 @@ static int mread(int fd, void* buf, size
        int rc;
        size_t cur = 0;
        struct timeval tv = {
-               .tv_sec = 0,
-               .tv_usec = NET_TIMEOUT * 1000
+               .tv_sec = HEARTBEAT_MS / 1000,
+               .tv_usec = (HEARTBEAT_MS % 1000) * 1000
        };
 
        if (!len)
@@ -649,8 +649,8 @@ static int mwrite(int fd, void* buf, siz
        size_t cur = 0;
        int rc;
        struct timeval tv = {
-               .tv_sec = 0,
-               .tv_usec = NET_TIMEOUT * 1000
+               .tv_sec = HEARTBEAT_MS / 1000,
+               .tv_usec = (HEARTBEAT_MS % 1000) * 1000
        };
 
        if (!len)
diff -r a4ddf47a1e91 -r a948403c8f99 tools/libxc/xc_domain_restore.c
--- a/tools/libxc/xc_domain_restore.c   Fri Feb 12 09:21:57 2010 +0000
+++ b/tools/libxc/xc_domain_restore.c   Fri Feb 12 09:23:10 2010 +0000
@@ -444,7 +444,7 @@ alloc_page:
 /* set when a consistent image is available */
 static int completed = 0;
 
-#define HEARTBEAT_MS 500
+#define HEARTBEAT_MS 1000
 
 #ifndef __MINIOS__
 static ssize_t read_exact_timed(int fd, void* buf, size_t size)
@@ -458,8 +458,8 @@ static ssize_t read_exact_timed(int fd, 
     {
         if ( completed ) {
             /* expect a heartbeat every HEARBEAT_MS ms maximum */
-            tv.tv_sec = 0;
-            tv.tv_usec = HEARTBEAT_MS * 1000;
+            tv.tv_sec = HEARTBEAT_MS / 1000;
+            tv.tv_usec = (HEARTBEAT_MS % 1000) * 1000;
 
             FD_ZERO(&rfds);
             FD_SET(fd, &rfds);

_______________________________________________
Xen-changelog mailing list
Xen-changelog@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-changelog


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.