aboutsummaryrefslogtreecommitdiff
path: root/include
diff options
context:
space:
mode:
authorNicholas Bellinger2017-01-19 15:45:57 -0800
committerNicholas Bellinger2017-02-26 16:21:06 -0800
commitc87ba9c49c1fa86261448b09c5f1b2223bf7efd9 (patch)
treed710dc3220024b8215cb2531fc55a122ebaba07c /include
parent17c61ad66f2e09a9014ab2d4e1f04c8294427db1 (diff)
target: Add counters for ABORT_TASK success + failure
This patch introduces two counters for ABORT_TASK success + failure under: /sys/kernel/config/target/core/$HBA/$DEV/statistics/scsi_tgt_dev/ that are useful for diagnosing various backend device latency and front fabric issues. Normally when folks see alot of aborts_complete happening, it means the backend device I/O completion latency is high, and not returning completions fast enough before host side timeouts trigger. And normally when folks see alot of aborts_no_task, it means completions are being posted by target-core into fabric driver code, but the responses aren't making it back to the host. Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
Diffstat (limited to 'include')
-rw-r--r--include/target/target_core_base.h2
1 files changed, 2 insertions, 0 deletions
diff --git a/include/target/target_core_base.h b/include/target/target_core_base.h
index 16d3be8395be..49ce5bc9912f 100644
--- a/include/target/target_core_base.h
+++ b/include/target/target_core_base.h
@@ -766,6 +766,8 @@ struct se_device {
u32 dev_index;
u64 creation_time;
atomic_long_t num_resets;
+ atomic_long_t aborts_complete;
+ atomic_long_t aborts_no_task;
atomic_long_t num_cmds;
atomic_long_t read_bytes;
atomic_long_t write_bytes;