Navigation

    Netgate Discussion Forum
    • Register
    • Login
    • Search
    • Categories
    • Recent
    • Tags
    • Popular
    • Users
    • Search

    Squid working, but getting some "operation timed out" errors

    pfSense Packages
    2
    2
    4963
    Loading More Posts
    • Oldest to Newest
    • Newest to Oldest
    • Most Votes
    Reply
    • Reply as topic
    Log in to reply
    This topic has been deleted. Only users with topic management privileges can see it.
    • E
      EmanuelG last edited by

      Hi people, I'm working now on the 1.0-BETA1-TESTING-SNAPSHOT-2-5-06, from an upgrade, and my squid was working almost right…

      I'm suffering from slow http surfing, and in some cases I got the "(60) Operation timed out".  I've checked the available bandwidth and it's about 60% available.

      I try disabling the traffic shapping completely, with no luck.  I decide to drop the captive portal couple weeks ago trying to find the reason for the slowdowns, but I can't figure it out  ???

      Is anybody having this behavior? I'm need someone gives me some light, or where to start find the problem.

      As always, thanks for this impressive piece of software :) guys!

      Emanuel Gonzalez
      Guatemala

      "I hear and I forget. I see and I remember. I do and I understand."
      Confucius

      1 Reply Last reply Reply Quote 0
      • C
        converter last edited by

        From the "Better Late Than Never" Department:

        Log into the box and use squidclient to query squid for stats:

        squidclient mgr:menu
        

        will give you a complete list of available commands.

        squidclient mgr:info
        

        is generally the first and most useful query to run.

        Look at the statistics related to median response time. Of particular interest will be the time it takes DNS to answer requests. Remember, squid is a proxy, that means that it's doing a lot of work in place of clients. Each request consumes system resources while it's being processed and the longer it takes to complete the transaction the longer those resources are unavailable to handle other requests.

        Also, make sure you're using diskd for async I/O, if possible, and that shared mem is tuned properly (if shared mem isn't tuned properly you'll get scary messages about resources being unavailable in the logs). NOTE: diskd can be a pain. If aufs works, use it instead. The last info I read on diskd v. aufs stated that in order for aufs to work on freebsd, threads would have to be able to do non-blocking I/O. I'm not sure about the current kernel, so more research is required.

        Don't use the default cache object removal policy, it's slow, inefficient and indiscriminate. Use one of the heap-based policies instead. Choose whichever policy meets your need, the primary distinguishing characteristic being their affinity for object size or age.

        Check the squid FAQ for more information.

        1 Reply Last reply Reply Quote 0
        • First post
          Last post