From 83aff95eb9d60aff5497e9f44a2ae906b86d8e88 Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Wed, 28 Nov 2012 12:28:24 -0800 Subject: libceph: remove 'osdtimeout' option This would reset a connection with any OSD that had an outstanding request that was taking more than N seconds. The idea was that if the OSD was buggy, the client could compensate by resending the request. In reality, this only served to hide server bugs, and we haven't actually seen such a bug in quite a while. Moreover, the userspace client code never did this. More importantly, often the request is taking a long time because the OSD is trying to recover, or overloaded, and killing the connection and retrying would only make the situation worse by giving the OSD more work to do. Signed-off-by: Sage Weil Reviewed-by: Alex Elder --- net/ceph/ceph_common.c | 3 +-- net/ceph/osd_client.c | 47 ++++------------------------------------------- 2 files changed, 5 insertions(+), 45 deletions(-) (limited to 'net') diff --git a/net/ceph/ceph_common.c b/net/ceph/ceph_common.c index a8020293f342..ee71ea26777a 100644 --- a/net/ceph/ceph_common.c +++ b/net/ceph/ceph_common.c @@ -305,7 +305,6 @@ ceph_parse_options(char *options, const char *dev_name, /* start with defaults */ opt->flags = CEPH_OPT_DEFAULT; - opt->osd_timeout = CEPH_OSD_TIMEOUT_DEFAULT; opt->osd_keepalive_timeout = CEPH_OSD_KEEPALIVE_DEFAULT; opt->mount_timeout = CEPH_MOUNT_TIMEOUT_DEFAULT; /* seconds */ opt->osd_idle_ttl = CEPH_OSD_IDLE_TTL_DEFAULT; /* seconds */ @@ -391,7 +390,7 @@ ceph_parse_options(char *options, const char *dev_name, /* misc */ case Opt_osdtimeout: - opt->osd_timeout = intval; + pr_warning("ignoring deprecated osdtimeout option\n"); break; case Opt_osdkeepalivetimeout: opt->osd_keepalive_timeout = intval; diff --git a/net/ceph/osd_client.c b/net/ceph/osd_client.c index ccbdfbba9e53..7ebfe13267e6 100644 --- a/net/ceph/osd_client.c +++ b/net/ceph/osd_client.c @@ -608,14 +608,6 @@ static void __kick_osd_requests(struct ceph_osd_client *osdc, } } -static void kick_osd_requests(struct ceph_osd_client *osdc, - struct ceph_osd *kickosd) -{ - mutex_lock(&osdc->request_mutex); - __kick_osd_requests(osdc, kickosd); - mutex_unlock(&osdc->request_mutex); -} - /* * If the osd connection drops, we need to resubmit all requests. */ @@ -629,7 +621,9 @@ static void osd_reset(struct ceph_connection *con) dout("osd_reset osd%d\n", osd->o_osd); osdc = osd->o_osdc; down_read(&osdc->map_sem); - kick_osd_requests(osdc, osd); + mutex_lock(&osdc->request_mutex); + __kick_osd_requests(osdc, osd); + mutex_unlock(&osdc->request_mutex); send_queued(osdc); up_read(&osdc->map_sem); } @@ -1091,12 +1085,10 @@ static void handle_timeout(struct work_struct *work) { struct ceph_osd_client *osdc = container_of(work, struct ceph_osd_client, timeout_work.work); - struct ceph_osd_request *req, *last_req = NULL; + struct ceph_osd_request *req; struct ceph_osd *osd; - unsigned long timeout = osdc->client->options->osd_timeout * HZ; unsigned long keepalive = osdc->client->options->osd_keepalive_timeout * HZ; - unsigned long last_stamp = 0; struct list_head slow_osds; dout("timeout\n"); down_read(&osdc->map_sem); @@ -1105,37 +1097,6 @@ static void handle_timeout(struct work_struct *work) mutex_lock(&osdc->request_mutex); - /* - * reset osds that appear to be _really_ unresponsive. this - * is a failsafe measure.. we really shouldn't be getting to - * this point if the system is working properly. the monitors - * should mark the osd as failed and we should find out about - * it from an updated osd map. - */ - while (timeout && !list_empty(&osdc->req_lru)) { - req = list_entry(osdc->req_lru.next, struct ceph_osd_request, - r_req_lru_item); - - /* hasn't been long enough since we sent it? */ - if (time_before(jiffies, req->r_stamp + timeout)) - break; - - /* hasn't been long enough since it was acked? */ - if (req->r_request->ack_stamp == 0 || - time_before(jiffies, req->r_request->ack_stamp + timeout)) - break; - - BUG_ON(req == last_req && req->r_stamp == last_stamp); - last_req = req; - last_stamp = req->r_stamp; - - osd = req->r_osd; - BUG_ON(!osd); - pr_warning(" tid %llu timed out on osd%d, will reset osd\n", - req->r_tid, osd->o_osd); - __kick_osd_requests(osdc, osd); - } - /* * ping osds that are a bit slow. this ensures that if there * is a break in the TCP connection we will notice, and reopen -- cgit v1.2.1