From 1588471f535f7de1e3e9f38afe9fced8c8d0e1ab Mon Sep 17 00:00:00 2001 From: Emoly Liu Date: Fri, 7 Dec 2012 09:56:34 +0800 Subject: [PATCH] LU-2433 lnet: enable asymmetric router failure detection Asynmmetric router failure detection is disabled by default, it's helpful to enable it by default. The original change is at https://github.com/chaos/lustre/ commit/872bfc41cc858a6c01cae3648e0ca329a414a436 Signed-off-by: Christopher J. Morrone Signed-off-by: Emoly Liu Change-Id: I259e00911cdf3d7caecb98c9832d40a91b571596 Reviewed-on: http://review.whamcloud.com/4753 Reviewed-by: Isaac Huang Reviewed-by: Liang Zhen Tested-by: Hudson Tested-by: Maloo Reviewed-by: Oleg Drokin --- lnet/lnet/router.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/lnet/lnet/router.c b/lnet/lnet/router.c index 5bb07dd..da0373b 100644 --- a/lnet/lnet/router.c +++ b/lnet/lnet/router.c @@ -82,23 +82,23 @@ lnet_peer_buffer_credits(lnet_ni_t *ni) static int check_routers_before_use = 0; CFS_MODULE_PARM(check_routers_before_use, "i", int, 0444, - "Assume routers are down and ping them before use"); + "Assume routers are down and ping them before use"); -static int avoid_asym_router_failure = 0; +static int avoid_asym_router_failure = 1; CFS_MODULE_PARM(avoid_asym_router_failure, "i", int, 0644, - "Avoid asymmetrical failures: reserved, use at your own risk"); + "Avoid asymmetrical router failures (0 to disable)"); -static int dead_router_check_interval = 0; +static int dead_router_check_interval = 60; CFS_MODULE_PARM(dead_router_check_interval, "i", int, 0644, - "Seconds between dead router health checks (<= 0 to disable)"); + "Seconds between dead router health checks (<= 0 to disable)"); -static int live_router_check_interval = 0; +static int live_router_check_interval = 60; CFS_MODULE_PARM(live_router_check_interval, "i", int, 0644, - "Seconds between live router health checks (<= 0 to disable)"); + "Seconds between live router health checks (<= 0 to disable)"); static int router_ping_timeout = 50; CFS_MODULE_PARM(router_ping_timeout, "i", int, 0644, - "Seconds to wait for the reply to a router health query"); + "Seconds to wait for the reply to a router health query"); int lnet_peers_start_down(void) -- 1.8.3.1