Christian Schwarz bf20d78292 fix(page_service): page reconstruct error log does not include shard_id label (#10680) ...

# Problem

Before this PR, the `shard_id` field was missing when page_service logs
a reconstruct error.

This was caused by batching-related refactorings.

Example from staging:

```
2025-01-30T07:10:04.346022Z ERROR page_service_conn_main{peer_addr=...}:process_query{tenant_id=... timeline_id=...}:handle_pagerequests:request:handle_get_page_at_lsn_request_batched{req_lsn=FFFFFFFF/FFFFFFFF}: error reading relation or page version: Read error: whole vectored get request failed because one or more of the requested keys were missing: could not find data for key  ...
```

# Changes

Delay creation of the handler-specific span until after shard routing

This also avoids the need for the record() call in the pagestream hot
path.

# Testing

Manual testing with a failpoint that is part of this PR's history but
will be squashed away.


# Refs

- fixes https://github.com/neondatabase/neon/issues/10599

2025-02-07 19:45:39 +00:00

..

benches

pageserver: coalesce index uploads when possible (#10248)

2025-01-14 21:10:17 +00:00

client

storcon: signal LSN wait to pageserver during live migration (#10452)

2025-01-28 17:33:07 +00:00

compaction

Update axum to 0.8.1 (#10332)

2025-01-28 15:32:59 +00:00

ctl

pageserver: add page_trace API for debugging (#10293)

2025-01-15 19:07:22 +00:00

pagebench

pageserver: issue concurrent IO on the read path (#9353)

2025-01-22 15:30:23 +00:00

src

fix(page_service): page reconstruct error log does not include shard_id label (#10680)

2025-02-07 19:45:39 +00:00

test_data

pageserver: add layer visibility calculation (#8511)

2024-08-01 09:25:35 +00:00

Cargo.toml

Use num_cpus from the workspace in pageserver (#10545)

2025-01-29 15:45:36 +00:00