diff options
| author | Dave Wysochanski <dwysocha@redhat.com> | 2019-05-23 23:13:50 +0300 |
|---|---|---|
| committer | Trond Myklebust <trond.myklebust@hammerspace.com> | 2019-07-06 21:54:51 +0300 |
| commit | a332518fda4731c07394164b3edcbb6efaf4c4d7 (patch) | |
| tree | ddb6ffaee7a3327fe10402914093ae8bdd1ca031 | |
| parent | 93ba048e1b9f95ac7e1584fa33ed4d985bf1133d (diff) | |
| download | linux-a332518fda4731c07394164b3edcbb6efaf4c4d7.tar.xz | |
SUNRPC: Count ops completing with tk_status < 0
We often see various error conditions with NFS4.x that show up with
a very high operation count all completing with tk_status < 0 in a
short period of time. Add a count to rpc_iostats to record on a
per-op basis the ops that complete in this manner, which will
enable lower overhead diagnostics.
Signed-off-by: Dave Wysochanski <dwysocha@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
| -rw-r--r-- | include/linux/sunrpc/metrics.h | 7 | ||||
| -rw-r--r-- | net/sunrpc/stats.c | 8 |
2 files changed, 12 insertions, 3 deletions
diff --git a/include/linux/sunrpc/metrics.h b/include/linux/sunrpc/metrics.h index 1b3751327575..0ee3f7052846 100644 --- a/include/linux/sunrpc/metrics.h +++ b/include/linux/sunrpc/metrics.h @@ -30,7 +30,7 @@ #include <linux/ktime.h> #include <linux/spinlock.h> -#define RPC_IOSTATS_VERS "1.0" +#define RPC_IOSTATS_VERS "1.1" struct rpc_iostats { spinlock_t om_lock; @@ -66,6 +66,11 @@ struct rpc_iostats { ktime_t om_queue, /* queued for xmit */ om_rtt, /* RPC RTT */ om_execute; /* RPC execution */ + /* + * The count of operations that complete with tk_status < 0. + * These statuses usually indicate error conditions. + */ + unsigned long om_error_status; } ____cacheline_aligned; struct rpc_task; diff --git a/net/sunrpc/stats.c b/net/sunrpc/stats.c index 2f7bde82450b..48ea776364f8 100644 --- a/net/sunrpc/stats.c +++ b/net/sunrpc/stats.c @@ -177,6 +177,8 @@ void rpc_count_iostats_metrics(const struct rpc_task *task, execute = ktime_sub(now, task->tk_start); op_metrics->om_execute = ktime_add(op_metrics->om_execute, execute); + if (task->tk_status < 0) + op_metrics->om_error_status++; spin_unlock(&op_metrics->om_lock); @@ -219,13 +221,14 @@ static void _add_rpc_iostats(struct rpc_iostats *a, struct rpc_iostats *b) a->om_queue = ktime_add(a->om_queue, b->om_queue); a->om_rtt = ktime_add(a->om_rtt, b->om_rtt); a->om_execute = ktime_add(a->om_execute, b->om_execute); + a->om_error_status += b->om_error_status; } static void _print_rpc_iostats(struct seq_file *seq, struct rpc_iostats *stats, int op, const struct rpc_procinfo *procs) { _print_name(seq, op, procs); - seq_printf(seq, "%lu %lu %lu %llu %llu %llu %llu %llu\n", + seq_printf(seq, "%lu %lu %lu %llu %llu %llu %llu %llu %lu\n", stats->om_ops, stats->om_ntrans, stats->om_timeouts, @@ -233,7 +236,8 @@ static void _print_rpc_iostats(struct seq_file *seq, struct rpc_iostats *stats, stats->om_bytes_recv, ktime_to_ms(stats->om_queue), ktime_to_ms(stats->om_rtt), - ktime_to_ms(stats->om_execute)); + ktime_to_ms(stats->om_execute), + stats->om_error_status); } void rpc_clnt_show_stats(struct seq_file *seq, struct rpc_clnt *clnt) |