fix(memory): increase reserved mem to 30%

[fuyufjh]

I hereby agree to the terms of the RisingWave Labs, Inc. Contributor License Agreement.

What's changed and what's your intention?

The reserved_memory is designed for the difference between actual usage (jemalloc_allocated) and process usage (process_resident_memory).

risingwave/src/compute/src/memory/config.rs

Lines 40 to 42 in b640bc9

	/// Each compute node reserves some memory for stack and code segment of processes, allocation
	/// overhead, network buffer, etc. based on `SYSTEM_RESERVED_MEMORY_PROPORTION`. The reserve memory
	/// size must be larger than `MIN_SYSTEM_RESERVED_MEMORY_MB`

As being observed in #13610,

• jemalloc_allocated_bytes: 8.59 GB
• jemalloc_active_bytes: 9.93 GB
• Node memory (RSS): 12.3 GB

The ratio is 8.59/12.3 = 0.698 ≈ 0.7.

Passed longevity test: https://buildkite.com/risingwave-test/longevity-test/builds/818 (Previously this would OOM for 2 times)

Checklist

• I have written necessary rustdoc comments
• I have added necessary unit tests and integration tests
• I have added test labels as necessary. See details.
• I have added fuzzing tests or opened an issue to track them. (Optional, recommended for new SQL features Sqlsmith: Sql feature generation #7934).
• My PR contains breaking changes. (If it deprecates some features, please create a tracking issue to remove them in the future).
• All checks passed in ./risedev check (or alias, ./risedev c)
• My PR changes performance-critical code. (Please run macro/micro-benchmarks and show the results.)

• My PR contains critical fixes that are necessary to be merged into the latest release. (Please check out the details)

Documentation

• My PR needs documentation updates. (Please use the Release note section below to summarize the impact on users)

Release note

If this PR includes changes that directly affect users or other significant modifications relevant to the community, kindly draft a release note to provide a concise summary of these changes. Please prioritize highlighting the impact these changes will have on users.

[KeXiangWang]

LGTM

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (a3ea7c0) 68.18% compared to head (c71a926) 68.18%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #13648      +/-   ##
==========================================
- Coverage   68.18%   68.18%   -0.01%     
==========================================
  Files        1521     1521              
  Lines      261709   261709              
==========================================
- Hits       178454   178442      -12     
- Misses      83255    83267      +12     

Flag	Coverage Δ
rust	68.18% <100.00%> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[wcy-fdu]

Previously I also made similar adjustments, hoping to pass the longevity test.
My change is:

const SYSTEM_RESERVED_MEMORY_PROPORTION: f64 = 0.35;
const STORAGE_MEMORY_PROPORTION: f64 = 0.15;

and the longevity test passed without limit rdkakfa fetch_queue_size
https://buildkite.com/risingwave-test/longevity-test/builds/787

Not sure whether should we reduce storage memory size in the meantime. cc @hzxa21

[hzxa21]

LGTM

[yuhao-su]

Should we consider using stats.mapped or stats.resident instead of allocated in this case?

[fuyufjh]

Should we consider using stats.mapped or stats.resident instead of allocated in this case?

stats.mapped or stats.resident is closer to the actual memory usage (rss), but similar with rss, it doesn't reflect the actual bytes allocated by the program. I would tend to use system rss if we really consider stats.allocated to be too inaccurate.

PS. At the beginning, I planned to replace the jemalloc_allocated to system rss, but I would like to try this "conservative" solution first.