From aaaba34576407857f6146ff6c330f06e63fb2bf2 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Tue, 18 Mar 2014 12:30:09 +0100 Subject: drbd: implement csums-after-crash-only Checksum based resync trades CPU cycles for network bandwidth, in situations where we expect much of the to-be-resynced blocks to be actually identical on both sides already. In a "network hickup" scenario, it won't help: all to-be-resynced blocks will typically be different. The use case is for the resync of *potentially* different blocks after crash recovery -- the crash recovery had marked larger areas (those covered by the activity log) as need-to-be-resynced, just in case. Most of those blocks will be identical. This option makes it possible to configure checksum based resync, but only actually use it for the first resync after primary crash. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- include/linux/drbd_limits.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/linux/drbd_limits.h') diff --git a/include/linux/drbd_limits.h b/include/linux/drbd_limits.h index 17e50bb0052..9d2df1d5141 100644 --- a/include/linux/drbd_limits.h +++ b/include/linux/drbd_limits.h @@ -214,6 +214,7 @@ #define DRBD_ALLOW_TWO_PRIMARIES_DEF 0 #define DRBD_ALWAYS_ASBP_DEF 0 #define DRBD_USE_RLE_DEF 1 +#define DRBD_CSUMS_AFTER_CRASH_ONLY_DEF 0 #define DRBD_AL_STRIPES_MIN 1 #define DRBD_AL_STRIPES_MAX 1024 -- cgit v1.2.3-70-g09d2 From 5d0b17f1a29e8189d04aef447a3a53cfd05529b2 Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Tue, 18 Mar 2014 14:24:35 +0100 Subject: drbd: New net configuration option socket-check-timeout In setups involving a DRBD-proxy and connections that experience a lot of buffer-bloat it might be necessary to set ping-timeout to an unusual high value. By default DRBD uses the same value to wait if a newly established TCP-connection is stable. Since the DRBD-proxy is usually located in the same data center such a long wait time may hinder DRBD's connect process. In such setups socket-check-timeout should be set to at least to the round trip time between DRBD and DRBD-proxy. I.e. in most cases to 1. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_receiver.c | 46 +++++++++++++++++++++++++------------- include/linux/drbd_genl.h | 1 + include/linux/drbd_limits.h | 5 +++++ 3 files changed, 36 insertions(+), 16 deletions(-) (limited to 'include/linux/drbd_limits.h') diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index 7da83f3a61e..b89e6fb468c 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -819,7 +819,7 @@ static int receive_first_packet(struct drbd_connection *connection, struct socke * drbd_socket_okay() - Free the socket if its connection is not okay * @sock: pointer to the pointer to the socket. */ -static int drbd_socket_okay(struct socket **sock) +static bool drbd_socket_okay(struct socket **sock) { int rr; char tb[4]; @@ -837,6 +837,30 @@ static int drbd_socket_okay(struct socket **sock) return false; } } + +static bool connection_established(struct drbd_connection *connection, + struct socket **sock1, + struct socket **sock2) +{ + struct net_conf *nc; + int timeout; + bool ok; + + if (!*sock1 || !*sock2) + return false; + + rcu_read_lock(); + nc = rcu_dereference(connection->net_conf); + timeout = (nc->sock_check_timeo ?: nc->ping_timeo) * HZ / 10; + rcu_read_unlock(); + schedule_timeout_interruptible(timeout); + + ok = drbd_socket_okay(sock1); + ok = drbd_socket_okay(sock2) && ok; + + return ok; +} + /* Gets called if a connection is established, or if a new minor gets created in a connection */ int drbd_connected(struct drbd_peer_device *peer_device) @@ -878,8 +902,8 @@ static int conn_connect(struct drbd_connection *connection) struct drbd_socket sock, msock; struct drbd_peer_device *peer_device; struct net_conf *nc; - int vnr, timeout, h, ok; - bool discard_my_data; + int vnr, timeout, h; + bool discard_my_data, ok; enum drbd_state_rv rv; struct accept_wait_data ad = { .connection = connection, @@ -923,17 +947,8 @@ static int conn_connect(struct drbd_connection *connection) } } - if (sock.socket && msock.socket) { - rcu_read_lock(); - nc = rcu_dereference(connection->net_conf); - timeout = nc->ping_timeo * HZ / 10; - rcu_read_unlock(); - schedule_timeout_interruptible(timeout); - ok = drbd_socket_okay(&sock.socket); - ok = drbd_socket_okay(&msock.socket) && ok; - if (ok) - break; - } + if (connection_established(connection, &sock.socket, &msock.socket)) + break; retry: s = drbd_wait_for_connect(connection, &ad); @@ -979,8 +994,7 @@ randomize: goto out_release_sockets; } - ok = drbd_socket_okay(&sock.socket); - ok = drbd_socket_okay(&msock.socket) && ok; + ok = connection_established(connection, &sock.socket, &msock.socket); } while (!ok); if (ad.s_listen) diff --git a/include/linux/drbd_genl.h b/include/linux/drbd_genl.h index 71fc924c53f..7b131ed8f9c 100644 --- a/include/linux/drbd_genl.h +++ b/include/linux/drbd_genl.h @@ -174,6 +174,7 @@ GENL_struct(DRBD_NLA_NET_CONF, 5, net_conf, /* 9: __str_field_def(31, DRBD_GENLA_F_MANDATORY, name, SHARED_SECRET_MAX) */ /* 9: __u32_field(32, DRBD_F_REQUIRED | DRBD_F_INVARIANT, peer_node_id) */ __flg_field_def(33, 0 /* OPTIONAL */, csums_after_crash_only, DRBD_CSUMS_AFTER_CRASH_ONLY_DEF) + __u32_field_def(34, 0 /* OPTIONAL */, sock_check_timeo, DRBD_SOCKET_CHECK_TIMEO_DEF) ) GENL_struct(DRBD_NLA_SET_ROLE_PARMS, 6, set_role_parms, diff --git a/include/linux/drbd_limits.h b/include/linux/drbd_limits.h index 9d2df1d5141..8ac8c5d9a3a 100644 --- a/include/linux/drbd_limits.h +++ b/include/linux/drbd_limits.h @@ -225,4 +225,9 @@ #define DRBD_AL_STRIPE_SIZE_MAX 16777216 #define DRBD_AL_STRIPE_SIZE_DEF 32 #define DRBD_AL_STRIPE_SIZE_SCALE 'k' /* kilobytes */ + +#define DRBD_SOCKET_CHECK_TIMEO_MIN 0 +#define DRBD_SOCKET_CHECK_TIMEO_MAX DRBD_PING_TIMEO_MAX +#define DRBD_SOCKET_CHECK_TIMEO_DEF 0 +#define DRBD_SOCKET_CHECK_TIMEO_SCALE '1' #endif -- cgit v1.2.3-70-g09d2