Skip to content

TLS Crypto Memory Leak ( New Relic ) #1866

Description

@miketheprogrammer

Currently tracked on nodejs/node-v0.x-archive#9064
Being that io.js is an answer to the 'slowness' of NodeJS, I was hoping we could get a look at this issue in
iojs. There is an existing patch PR on the ticket.

Thank you for considering this. The memory leak is fairly large and serious for Micro Service oriented architectures especially.

Activity

  1. mscdex commented on Jun 1, 2015

    @mscdex
    Contributor

    @miketheprogrammer Have you tried with recent versions of io.js (v1.8.2+)? I know @indutny added some TLS memory leak fixes not too long ago...

  2. added
    tlsIssues and PRs related to the tls subsystem.
    on Jun 1, 2015
  3. miketheprogrammer commented on Jun 1, 2015

    @miketheprogrammer
    Author

    I will definitely attempt using iojs latest on one of our production servers. However I was holding of on upgrading for a little while and staying on 0.12.

    I feel from reading it that they are two seperate issues, however maybe this PR will at the very least alleviate some of the issues with the NewRelic memory leak. Imagine 800mb of extra memory over 6-10 hours depending on server activity. Absurd.

  4. ChALkeR commented on Jun 1, 2015

    @ChALkeR
    Member

    So, is this report against 0.12? What's the point of filing an issue at this moment, if you haven't tested iojs, then?

    You could read this: #1522, it was also related with a lot of cert copies (btw that was not technically a leak), but there are no guarantees that you issue is the same. PR: #1529.

  5. rvagg commented on Jun 2, 2015

    @rvagg
    Member

    Not to be rude, but I'm going to close this issue @miketheprogrammer because the delta in tls between the current io.js and 0.12 is very large, almost everything has been touched, including the version of OpenSSL. So I'd suggest that this is an issue for 0.12 unless you've tested with io.js >= 2.0 and have some numbers to report, preferably with a piece of code that can be used to reproduce the issue.

    Feel free to suggest this be re-opened if you think I've read this wrong.

  6. miketheprogrammer commented on Jun 2, 2015

    @miketheprogrammer
    Author

    It's okay, I will test first. Apologies.

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

Metadata

Metadata

Assignees

No one assigned

    Labels

    invalidIssues and PRs that are invalid.tlsIssues and PRs related to the tls subsystem.

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions