mirrors/scylladb
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

alternator: track per-table latency for batch get/write operations

Browse Source 
Batch operations were updating only global latency histograms, which left
table-level latency metrics incomplete.

This change computes request duration once at the end of each operation and
reuses it to update both global and per-table latency stats:

Latencies are stored per table used,

This aligns batch read/write metric behavior with other operations and improves
per-table observability.

Signed-off-by: Amnon Heiman <amnon@scylladb.com>
This commit is contained in:
Amnon Heiman
2026-02-19 20:58:39 +02:00
parent 9dff9752b4
commit 29e0b4e08c
1 changed files with 11 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
alternator/executor.cc
View File

@@ -3464,7 +3464,11 @@ future<executor::request_return_type> executor::batch_write_item(client_state& c
if (should_add_wcu) {
rjson::add(ret, "ConsumedCapacity", std::move(consumed_capacity));
}
_stats.api_operations.batch_write_item_latency.mark(std::chrono::steady_clock::now() - start_time);
auto duration = std::chrono::steady_clock::now() - start_time;
_stats.api_operations.batch_write_item_latency.mark(duration);
for (const auto& w : per_table_wcu) {
w.first->api_operations.batch_write_item_latency.mark(duration);
}
co_return rjson::print(std::move(ret));
}
@@ -4975,7 +4979,12 @@ future<executor::request_return_type> executor::batch_get_item(client_state& cli
if (!some_succeeded && eptr) {
co_await coroutine::return_exception_ptr(std::move(eptr));
}
_stats.api_operations.batch_get_item_latency.mark(std::chrono::steady_clock::now() - start_time);
auto duration = std::chrono::steady_clock::now() - start_time;
_stats.api_operations.batch_get_item_latency.mark(duration);
for (const table_requests& rs : requests) {
lw_shared_ptr<stats> per_table_stats = get_stats_from_schema(_proxy, *rs.schema);
per_table_stats->api_operations.batch_get_item_latency.mark(duration);
}
if (is_big(response)) {
co_return make_streamed(std::move(response));
} else {