Skip to content

Add cache stripe lock contention metric #12839

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
bryancall merged 3 commits into from
Feb 2, 2026
+95 −71
Merged

Add cache stripe lock contention metric #12839

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
5ec8441
Add stripe lock contention metric
bryancall Jan 29, 2026
4bde886
Fix stripe lock contention metric accuracy
bryancall Jan 29, 2026
f488fe8
Add writer lock contention metric
bryancall Jan 29, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
72 changes: 37 additions & 35 deletions src/iocore/cache/CacheProcessor.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1115,41 +1115,43 @@ register_cache_stats(CacheStatsBlock *rsb, const std::string &prefix)
rsb->fragment_document_count[2] = ts::Metrics::Counter::createPtr(prefix + ".frags_per_doc.3+");

// And then everything else
rsb->bytes_used = ts::Metrics::Gauge::createPtr(prefix + ".bytes_used");
rsb->bytes_total = ts::Metrics::Gauge::createPtr(prefix + ".bytes_total");
rsb->stripes = ts::Metrics::Gauge::createPtr(prefix + ".stripes");
rsb->ram_cache_bytes_total = ts::Metrics::Gauge::createPtr(prefix + ".ram_cache.total_bytes");
rsb->ram_cache_bytes = ts::Metrics::Gauge::createPtr(prefix + ".ram_cache.bytes_used");
rsb->ram_cache_hits = ts::Metrics::Counter::createPtr(prefix + ".ram_cache.hits");
rsb->last_open_read_hits = ts::Metrics::Counter::createPtr(prefix + ".last_open_read.hits");
rsb->agg_buffer_hits = ts::Metrics::Counter::createPtr(prefix + ".aggregation_buffer.hits");
rsb->ram_cache_misses = ts::Metrics::Counter::createPtr(prefix + ".ram_cache.misses");
rsb->all_mem_misses = ts::Metrics::Counter::createPtr(prefix + ".all_memory_caches.misses");
rsb->pread_count = ts::Metrics::Counter::createPtr(prefix + ".pread_count");
rsb->percent_full = ts::Metrics::Gauge::createPtr(prefix + ".percent_full");
rsb->read_seek_fail = ts::Metrics::Counter::createPtr(prefix + ".read.seek.failure");
rsb->read_invalid = ts::Metrics::Counter::createPtr(prefix + ".read.invalid");
rsb->write_backlog_failure = ts::Metrics::Counter::createPtr(prefix + ".write.backlog.failure");
rsb->direntries_total = ts::Metrics::Gauge::createPtr(prefix + ".direntries.total");
rsb->direntries_used = ts::Metrics::Gauge::createPtr(prefix + ".direntries.used");
rsb->directory_collision = ts::Metrics::Counter::createPtr(prefix + ".directory_collision");
rsb->read_busy_success = ts::Metrics::Counter::createPtr(prefix + ".read_busy.success");
rsb->read_busy_failure = ts::Metrics::Counter::createPtr(prefix + ".read_busy.failure");
rsb->write_bytes = ts::Metrics::Counter::createPtr(prefix + ".write_bytes_stat");
rsb->hdr_vector_marshal = ts::Metrics::Counter::createPtr(prefix + ".vector_marshals");
rsb->hdr_marshal = ts::Metrics::Counter::createPtr(prefix + ".hdr_marshals");
rsb->hdr_marshal_bytes = ts::Metrics::Counter::createPtr(prefix + ".hdr_marshal_bytes");
rsb->gc_bytes_evacuated = ts::Metrics::Counter::createPtr(prefix + ".gc_bytes_evacuated");
rsb->gc_frags_evacuated = ts::Metrics::Counter::createPtr(prefix + ".gc_frags_evacuated");
rsb->directory_wrap = ts::Metrics::Counter::createPtr(prefix + ".wrap_count");
rsb->directory_sync_count = ts::Metrics::Counter::createPtr(prefix + ".sync.count");
rsb->directory_sync_bytes = ts::Metrics::Counter::createPtr(prefix + ".sync.bytes");
rsb->directory_sync_time = ts::Metrics::Counter::createPtr(prefix + ".sync.time");
rsb->span_errors_read = ts::Metrics::Counter::createPtr(prefix + ".span.errors.read");
rsb->span_errors_write = ts::Metrics::Counter::createPtr(prefix + ".span.errors.write");
rsb->span_failing = ts::Metrics::Gauge::createPtr(prefix + ".span.failing");
rsb->span_offline = ts::Metrics::Gauge::createPtr(prefix + ".span.offline");
rsb->span_online = ts::Metrics::Gauge::createPtr(prefix + ".span.online");
rsb->bytes_used = ts::Metrics::Gauge::createPtr(prefix + ".bytes_used");
rsb->bytes_total = ts::Metrics::Gauge::createPtr(prefix + ".bytes_total");
rsb->stripes = ts::Metrics::Gauge::createPtr(prefix + ".stripes");
rsb->ram_cache_bytes_total = ts::Metrics::Gauge::createPtr(prefix + ".ram_cache.total_bytes");
rsb->ram_cache_bytes = ts::Metrics::Gauge::createPtr(prefix + ".ram_cache.bytes_used");
rsb->ram_cache_hits = ts::Metrics::Counter::createPtr(prefix + ".ram_cache.hits");
rsb->last_open_read_hits = ts::Metrics::Counter::createPtr(prefix + ".last_open_read.hits");
rsb->agg_buffer_hits = ts::Metrics::Counter::createPtr(prefix + ".aggregation_buffer.hits");
rsb->ram_cache_misses = ts::Metrics::Counter::createPtr(prefix + ".ram_cache.misses");
rsb->all_mem_misses = ts::Metrics::Counter::createPtr(prefix + ".all_memory_caches.misses");
rsb->pread_count = ts::Metrics::Counter::createPtr(prefix + ".pread_count");
rsb->percent_full = ts::Metrics::Gauge::createPtr(prefix + ".percent_full");
rsb->read_seek_fail = ts::Metrics::Counter::createPtr(prefix + ".read.seek.failure");
rsb->read_invalid = ts::Metrics::Counter::createPtr(prefix + ".read.invalid");
rsb->write_backlog_failure = ts::Metrics::Counter::createPtr(prefix + ".write.backlog.failure");
rsb->direntries_total = ts::Metrics::Gauge::createPtr(prefix + ".direntries.total");
rsb->direntries_used = ts::Metrics::Gauge::createPtr(prefix + ".direntries.used");
rsb->directory_collision = ts::Metrics::Counter::createPtr(prefix + ".directory_collision");
rsb->read_busy_success = ts::Metrics::Counter::createPtr(prefix + ".read_busy.success");
rsb->read_busy_failure = ts::Metrics::Counter::createPtr(prefix + ".read_busy.failure");
rsb->write_bytes = ts::Metrics::Counter::createPtr(prefix + ".write_bytes_stat");
rsb->hdr_vector_marshal = ts::Metrics::Counter::createPtr(prefix + ".vector_marshals");
rsb->hdr_marshal = ts::Metrics::Counter::createPtr(prefix + ".hdr_marshals");
rsb->hdr_marshal_bytes = ts::Metrics::Counter::createPtr(prefix + ".hdr_marshal_bytes");
rsb->gc_bytes_evacuated = ts::Metrics::Counter::createPtr(prefix + ".gc_bytes_evacuated");
rsb->gc_frags_evacuated = ts::Metrics::Counter::createPtr(prefix + ".gc_frags_evacuated");
rsb->directory_wrap = ts::Metrics::Counter::createPtr(prefix + ".wrap_count");
rsb->directory_sync_count = ts::Metrics::Counter::createPtr(prefix + ".sync.count");
rsb->directory_sync_bytes = ts::Metrics::Counter::createPtr(prefix + ".sync.bytes");
rsb->directory_sync_time = ts::Metrics::Counter::createPtr(prefix + ".sync.time");
rsb->span_errors_read = ts::Metrics::Counter::createPtr(prefix + ".span.errors.read");
rsb->span_errors_write = ts::Metrics::Counter::createPtr(prefix + ".span.errors.write");
rsb->span_failing = ts::Metrics::Gauge::createPtr(prefix + ".span.failing");
rsb->span_offline = ts::Metrics::Gauge::createPtr(prefix + ".span.offline");
rsb->span_online = ts::Metrics::Gauge::createPtr(prefix + ".span.online");
rsb->stripe_lock_contention = ts::Metrics::Counter::createPtr(prefix + ".stripe.lock_contention");
rsb->writer_lock_contention = ts::Metrics::Counter::createPtr(prefix + ".writer.lock_contention");
}

void
Expand Down
2 changes: 1 addition & 1 deletion src/iocore/cache/CacheRead.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -275,7 +275,7 @@ CacheVC::openReadFromWriter(int event, Event *e)
CACHE_TRY_LOCK(writer_lock, write_vc->mutex, mutex->thread_holding);
if (!writer_lock.is_locked()) {
DDbg(dbg_ctl_cache_read_agg, "%p: key: %X lock miss", this, first_key.slice32(1));
VC_SCHED_LOCK_RETRY();
VC_SCHED_WRITER_LOCK_RETRY();
}
MUTEX_RELEASE(lock);

Expand Down
20 changes: 20 additions & 0 deletions src/iocore/cache/P_CacheInternal.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -61,12 +61,28 @@ struct EvacuationBlock;

#define VC_LOCK_RETRY_EVENT() \
do { \
ts::Metrics::Counter::increment(cache_rsb.stripe_lock_contention); \
if (stripe && stripe->cache_vol) { \
ts::Metrics::Counter::increment(stripe->cache_vol->vol_rsb.stripe_lock_contention); \
} \
trigger = mutex->thread_holding->schedule_in_local(this, HRTIME_MSECONDS(cache_config_mutex_retry_delay), event); \
return EVENT_CONT; \
} while (0)

#define VC_SCHED_LOCK_RETRY() \
do { \
ts::Metrics::Counter::increment(cache_rsb.stripe_lock_contention); \
if (stripe && stripe->cache_vol) { \
ts::Metrics::Counter::increment(stripe->cache_vol->vol_rsb.stripe_lock_contention); \
} \
trigger = mutex->thread_holding->schedule_in_local(this, HRTIME_MSECONDS(cache_config_mutex_retry_delay)); \
return EVENT_CONT; \
} while (0)

// Variant for writer lock contention (write_vc->mutex during read aggregation)
#define VC_SCHED_WRITER_LOCK_RETRY() \
do { \
ts::Metrics::Counter::increment(cache_rsb.writer_lock_contention); \
trigger = mutex->thread_holding->schedule_in_local(this, HRTIME_MSECONDS(cache_config_mutex_retry_delay)); \
return EVENT_CONT; \
} while (0)
Expand Down Expand Up @@ -330,6 +346,10 @@ CacheVC::handleWriteLock(int /* event ATS_UNUSED */, Event *e)
{
CACHE_TRY_LOCK(lock, stripe->mutex, mutex->thread_holding);
if (!lock.is_locked()) {
ts::Metrics::Counter::increment(cache_rsb.stripe_lock_contention);
if (stripe && stripe->cache_vol) {
ts::Metrics::Counter::increment(stripe->cache_vol->vol_rsb.stripe_lock_contention);
}
set_agg_write_in_progress();
trigger = mutex->thread_holding->schedule_in_local(this, HRTIME_MSECONDS(cache_config_mutex_retry_delay));
return EVENT_CONT;
Expand Down
72 changes: 37 additions & 35 deletions src/iocore/cache/P_CacheStats.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,39 +37,41 @@ struct CacheStatsBlock {

ts::Metrics::Counter::AtomicType *fragment_document_count[3] = {nullptr, nullptr, nullptr}; // For 1, 2 and 3+ fragments

ts::Metrics::Gauge::AtomicType *bytes_used = nullptr;
ts::Metrics::Gauge::AtomicType *bytes_total = nullptr;
ts::Metrics::Gauge::AtomicType *stripes = nullptr;
ts::Metrics::Gauge::AtomicType *ram_cache_bytes = nullptr;
ts::Metrics::Gauge::AtomicType *ram_cache_bytes_total = nullptr;
ts::Metrics::Gauge::AtomicType *direntries_total = nullptr;
ts::Metrics::Gauge::AtomicType *direntries_used = nullptr;
ts::Metrics::Counter::AtomicType *ram_cache_hits = nullptr;
ts::Metrics::Counter::AtomicType *last_open_read_hits = nullptr;
ts::Metrics::Counter::AtomicType *agg_buffer_hits = nullptr;
ts::Metrics::Counter::AtomicType *ram_cache_misses = nullptr;
ts::Metrics::Counter::AtomicType *all_mem_misses = nullptr;
ts::Metrics::Counter::AtomicType *pread_count = nullptr;
ts::Metrics::Gauge::AtomicType *percent_full = nullptr;
ts::Metrics::Counter::AtomicType *read_seek_fail = nullptr;
ts::Metrics::Counter::AtomicType *read_invalid = nullptr;
ts::Metrics::Counter::AtomicType *write_backlog_failure = nullptr;
ts::Metrics::Counter::AtomicType *directory_collision = nullptr;
ts::Metrics::Counter::AtomicType *read_busy_success = nullptr;
ts::Metrics::Counter::AtomicType *read_busy_failure = nullptr;
ts::Metrics::Counter::AtomicType *gc_bytes_evacuated = nullptr;
ts::Metrics::Counter::AtomicType *gc_frags_evacuated = nullptr;
ts::Metrics::Counter::AtomicType *write_bytes = nullptr;
ts::Metrics::Counter::AtomicType *hdr_vector_marshal = nullptr;
ts::Metrics::Counter::AtomicType *hdr_marshal = nullptr;
ts::Metrics::Counter::AtomicType *hdr_marshal_bytes = nullptr;
ts::Metrics::Counter::AtomicType *directory_wrap = nullptr;
ts::Metrics::Counter::AtomicType *directory_sync_count = nullptr;
ts::Metrics::Counter::AtomicType *directory_sync_time = nullptr;
ts::Metrics::Counter::AtomicType *directory_sync_bytes = nullptr;
ts::Metrics::Counter::AtomicType *span_errors_read = nullptr;
ts::Metrics::Counter::AtomicType *span_errors_write = nullptr;
ts::Metrics::Gauge::AtomicType *span_offline = nullptr;
ts::Metrics::Gauge::AtomicType *span_online = nullptr;
ts::Metrics::Gauge::AtomicType *span_failing = nullptr;
ts::Metrics::Gauge::AtomicType *bytes_used = nullptr;
ts::Metrics::Gauge::AtomicType *bytes_total = nullptr;
ts::Metrics::Gauge::AtomicType *stripes = nullptr;
ts::Metrics::Gauge::AtomicType *ram_cache_bytes = nullptr;
ts::Metrics::Gauge::AtomicType *ram_cache_bytes_total = nullptr;
ts::Metrics::Gauge::AtomicType *direntries_total = nullptr;
ts::Metrics::Gauge::AtomicType *direntries_used = nullptr;
ts::Metrics::Counter::AtomicType *ram_cache_hits = nullptr;
ts::Metrics::Counter::AtomicType *last_open_read_hits = nullptr;
ts::Metrics::Counter::AtomicType *agg_buffer_hits = nullptr;
ts::Metrics::Counter::AtomicType *ram_cache_misses = nullptr;
ts::Metrics::Counter::AtomicType *all_mem_misses = nullptr;
ts::Metrics::Counter::AtomicType *pread_count = nullptr;
ts::Metrics::Gauge::AtomicType *percent_full = nullptr;
ts::Metrics::Counter::AtomicType *read_seek_fail = nullptr;
ts::Metrics::Counter::AtomicType *read_invalid = nullptr;
ts::Metrics::Counter::AtomicType *write_backlog_failure = nullptr;
ts::Metrics::Counter::AtomicType *directory_collision = nullptr;
ts::Metrics::Counter::AtomicType *read_busy_success = nullptr;
ts::Metrics::Counter::AtomicType *read_busy_failure = nullptr;
ts::Metrics::Counter::AtomicType *gc_bytes_evacuated = nullptr;
ts::Metrics::Counter::AtomicType *gc_frags_evacuated = nullptr;
ts::Metrics::Counter::AtomicType *write_bytes = nullptr;
ts::Metrics::Counter::AtomicType *hdr_vector_marshal = nullptr;
ts::Metrics::Counter::AtomicType *hdr_marshal = nullptr;
ts::Metrics::Counter::AtomicType *hdr_marshal_bytes = nullptr;
ts::Metrics::Counter::AtomicType *directory_wrap = nullptr;
ts::Metrics::Counter::AtomicType *directory_sync_count = nullptr;
ts::Metrics::Counter::AtomicType *directory_sync_time = nullptr;
ts::Metrics::Counter::AtomicType *directory_sync_bytes = nullptr;
ts::Metrics::Counter::AtomicType *span_errors_read = nullptr;
ts::Metrics::Counter::AtomicType *span_errors_write = nullptr;
ts::Metrics::Gauge::AtomicType *span_offline = nullptr;
ts::Metrics::Gauge::AtomicType *span_online = nullptr;
ts::Metrics::Gauge::AtomicType *span_failing = nullptr;
ts::Metrics::Counter::AtomicType *stripe_lock_contention = nullptr;
ts::Metrics::Counter::AtomicType *writer_lock_contention = nullptr;
};