From 0e04dc26cc594d31ee6b1382b452b6bc83b57937 Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Wed, 20 Jan 2016 17:50:31 +0100 Subject: [PATCH] libceph: pick a different monitor when reconnecting Don't try to reconnect to the same monitor when we fail to establish a session within a timeout or it's lost. For that, pick_new_mon() needs to see the old value of cur_mon, so don't clear it in __close_session() - all calls to __close_session() but one are followed by __open_session() anyway. __open_session() is only called when a new session needs to be established, so the "already open?" branch, which is now in the way, is simply dropped. Signed-off-by: Ilya Dryomov --- net/ceph/mon_client.c | 85 ++++++++++++++++++++++++++++++++++----------------- 1 file changed, 57 insertions(+), 28 deletions(-) diff --git a/net/ceph/mon_client.c b/net/ceph/mon_client.c index 89029916..accfded 100644 --- a/net/ceph/mon_client.c +++ b/net/ceph/mon_client.c @@ -122,45 +122,74 @@ static void __close_session(struct ceph_mon_client *monc) ceph_msg_revoke(monc->m_subscribe); ceph_msg_revoke_incoming(monc->m_subscribe_ack); ceph_con_close(&monc->con); - monc->cur_mon = -1; + monc->pending_auth = 0; ceph_auth_reset(monc->auth); } /* - * Open a session with a (new) monitor. + * Pick a new monitor at random and set cur_mon. If we are repicking + * (i.e. cur_mon is already set), be sure to pick a different one. */ -static int __open_session(struct ceph_mon_client *monc) +static void pick_new_mon(struct ceph_mon_client *monc) { - char r; - int ret; + int old_mon = monc->cur_mon; - if (monc->cur_mon < 0) { - get_random_bytes(&r, 1); - monc->cur_mon = r % monc->monmap->num_mon; - dout("open_session num=%d r=%d -> mon%d\n", - monc->monmap->num_mon, r, monc->cur_mon); - monc->sub_renew_after = jiffies; /* i.e., expired */ - monc->sub_renew_sent = 0; + BUG_ON(monc->monmap->num_mon < 1); - dout("open_session mon%d opening\n", monc->cur_mon); - ceph_con_open(&monc->con, - CEPH_ENTITY_TYPE_MON, monc->cur_mon, - &monc->monmap->mon_inst[monc->cur_mon].addr); + if (monc->monmap->num_mon == 1) { + monc->cur_mon = 0; + } else { + int max = monc->monmap->num_mon; + int o = -1; + int n; + + if (monc->cur_mon >= 0) { + if (monc->cur_mon < monc->monmap->num_mon) + o = monc->cur_mon; + if (o >= 0) + max--; + } - /* send an initial keepalive to ensure our timestamp is - * valid by the time we are in an OPENED state */ - ceph_con_keepalive(&monc->con); + n = prandom_u32() % max; + if (o >= 0 && n >= o) + n++; - /* initiatiate authentication handshake */ - ret = ceph_auth_build_hello(monc->auth, - monc->m_auth->front.iov_base, - monc->m_auth->front_alloc_len); - __send_prepared_auth_request(monc, ret); - } else { - dout("open_session mon%d already open\n", monc->cur_mon); + monc->cur_mon = n; } - return 0; + + dout("%s mon%d -> mon%d out of %d mons\n", __func__, old_mon, + monc->cur_mon, monc->monmap->num_mon); +} + +/* + * Open a session with a new monitor. + */ +static void __open_session(struct ceph_mon_client *monc) +{ + int ret; + + pick_new_mon(monc); + + monc->sub_renew_after = jiffies; /* i.e., expired */ + monc->sub_renew_sent = 0; + + dout("%s opening mon%d\n", __func__, monc->cur_mon); + ceph_con_open(&monc->con, CEPH_ENTITY_TYPE_MON, monc->cur_mon, + &monc->monmap->mon_inst[monc->cur_mon].addr); + + /* + * send an initial keepalive to ensure our timestamp is valid + * by the time we are in an OPENED state + */ + ceph_con_keepalive(&monc->con); + + /* initiate authentication handshake */ + ret = ceph_auth_build_hello(monc->auth, + monc->m_auth->front.iov_base, + monc->m_auth->front_alloc_len); + BUG_ON(ret <= 0); + __send_prepared_auth_request(monc, ret); } static bool __sub_expired(struct ceph_mon_client *monc) @@ -907,7 +936,7 @@ void ceph_monc_stop(struct ceph_mon_client *monc) mutex_lock(&monc->mutex); __close_session(monc); - + monc->cur_mon = -1; mutex_unlock(&monc->mutex); /* -- 2.7.4