Skip to content

Run apache2 under jemalloc#7858

Draft
legoktm wants to merge 1 commit into
developfrom
stg-jemalloc
Draft

Run apache2 under jemalloc#7858
legoktm wants to merge 1 commit into
developfrom
stg-jemalloc

Conversation

@legoktm

@legoktm legoktm commented Jun 9, 2026

Copy link
Copy Markdown
Member

jemalloc ends up resolving the memory fragmentation issues we've seen under APIv2 with the default glibc allocator.

It should be relatively safe to swap in given how battle tested jemalloc is by much larger corporate users.

See freedomofpress/securedrop-client#3234 (comment) and following comments.

Fixes #7817.

Test plan

  • build debs and install on production hardware
  • point the Inbox at said server and sync
  • observe that memory usage is low/normal (below some X%?)

Checklist

This change accounts for:

  • any required additional documentation
  • any necessary AppArmor changes (added or removed application files)
  • any impact on new SecureDrop installs and upgrades
  • our dependency update policy

@legoktm legoktm moved this to In Progress in SecureDrop Jun 9, 2026
jemalloc ends up resolving the memory fragmentation issues we've seen
under APIv2 with the default glibc allocator.

It should be relatively safe to swap in given how battle tested jemalloc
is by much larger corporate users.

See <freedomofpress/securedrop-client#3234 (comment)>
and following comments.

Fixes #7817.
@legoktm

legoktm commented Jun 10, 2026

Copy link
Copy Markdown
Member Author

I'm not seeing the same gains (or any gains whatsoever tbh) from using jemalloc on the server, maybe I misconfigured something this time around, or possibly the initial test wasn't accurate? I'll have to spend some more time on this.

@cfm

cfm commented Jun 12, 2026

Copy link
Copy Markdown
Member

If #7862 is right, there may still be a higher-level optimization to pursue....

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

Status: In Progress

Development

Successfully merging this pull request may close these issues.

APIv2: options for memory-bound scaling

2 participants