Memory leak when metrics endpoint is enabled, but not scraped

[mikroskeem]

Describe the bug

When metrics endpoint is enabled with e.g. --metrics=127.0.0.1:2112 flag, memory usage keeps growing if endpoint is not being actively scraped.

Steps to reproduce

1. Start node with --metrics flag, e.g. --metrics=127.0.0.1:2112
2. Leave it running for a few days
3. See Reth memory usage grow significantly (up to ~20GiB RSS in 3-4 days)
4. Scrape endpoint (e.g. via curl, or start Prometheus) and see endpoint taking long time to generate a response
5. See Reth memory usage drop

Node logs

N/A

Platform(s)

Linux (x86)

What version/commit are you on?

reth Version: 1.1.0
Commit SHA: 1ba631b
Build Timestamp: 2024-10-10T18:21:22.096485546Z
Build Features: asm_keccak,jemalloc
Build Profile: maxperf

What database version are you on?

Current database version: 2
Local database version: 2

Which chain / network are you on?

mainnet

What type of node are you running?

Full via --full flag

What prune config do you use, if any?

No response

If you've built Reth from source, provide the full command you used

No response

Code of Conduct

• I agree to follow the Code of Conduct