public inbox for development@lists.ipfire.org
 help / color / mirror / Atom feed
From: Matthias Fischer <matthias.fischer@ipfire.org>
To: development@lists.ipfire.org
Subject: Re: Problems with "Enable some performance tuning" => extremly slow downloads
Date: Fri, 22 Feb 2019 18:40:50 +0100	[thread overview]
Message-ID: <3271e163-ba27-be1e-8501-bc3befdf541a@ipfire.org> (raw)
In-Reply-To: <E2D5D010-AC5C-41B0-8F5E-37D58FF56A3D@ipfire.org>

[-- Attachment #1: Type: text/plain, Size: 2388 bytes --]

Hi,

On 21.02.2019 10:33, Michael Tremer wrote:
> On 21 Feb 2019, at 00:36, Matthias Fischer <matthias.fischer(a)ipfire.org> wrote:
>> 
>> Hi,
>> 
>> On 20.02.2019 16:40, Michael Tremer wrote:
>>> Interesting… These settings shouldn’t have any impact on any connections going through the firewall.
>> ...
>> ...
>>> Can you narrow it down to one specific setting of these by disabling one by one?
>> ...

Yes. And I finished testing:

I would say: "case can be closed".

I wasn't able to track this down to one of the tuned parameters of the
commit cited below. Must be another reason.

I activated options one by one and tested with simple 1GB / 5GB download
files from Netcologne and Hetzner.

Now all parameters are active again => downloads still run at normal
speed. This means 7.0MB/sec - with peaks at 7.6MB/sec - which is
absolutely ok for our line.

Mostly it happened in the evening or at night. I'll see what happens then.

Best,
Matthias

>> 
>>> -Michael
>>> 
>>>> On 20 Feb 2019, at 10:18, Matthias Fischer <matthias.fischer(a)ipfire.org> wrote:
>>>> 
>>>> Hi,
>>>> 
>>>> being curious, I tested commit
>>>> https://git.ipfire.org/?p=ipfire-2.x.git;a=commit;h=d03916e55851a243594ebf6f0c20c8f6d9092277
>>>> on my Core 127 / 32bit IPFire.
>>>> 
>>>> At first I didn't notice any differences, system was running as usual.
>>>> No important performance impact or change.
>>>> 
>>>> But yesterday, while starting some bigger downloads and closely
>>>> watching, I noticed that everytime someone started to download a
>>>> somewhat bigger file, e.g. 250-800 MB, downloading rates went down to a
>>>> crawl. Some downloads even aborted and nearly all where amazingly slow
>>>> (~150KB/s, normal: ~6.5 MB/s).
>>>> 
>>>> Restarting our Fritzbox and IPFire itself didn't help, all downloads
>>>> stayed that way.
>>>> 
>>>> After reverting the above commit in '/etc/sysctl.conf' and running
>>>> 'sysctl -p', system is running at full speed again: VDSL, 50Mbit down /
>>>> 10Mbit up.
>>>> 
>>>> Configuration:
>>>> Duo Box with Core 127/32bit. Running 'privoxy 3.0.28', 'squid 4.6'
>>>> (non-transparent, 512 MB RAM only), 'squidguard 1.5 beta',
>>>> 'squidclamav', 'snort / guardian', 'unbound 1.9.0' with DoT/TFO.
>>>> 
>>>> Could someone please test and confirm (or not ;-) ).
>>>> 
>>>> Best,
>>>> Matthias

      parent reply	other threads:[~2019-02-22 17:40 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-02-20 10:18 Matthias Fischer
2019-02-20 15:40 ` Michael Tremer
2019-02-21  0:36   ` Matthias Fischer
2019-02-21  9:33     ` Michael Tremer
2019-02-21 15:41       ` Matthias Fischer
2019-02-22 17:40       ` Matthias Fischer [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=3271e163-ba27-be1e-8501-bc3befdf541a@ipfire.org \
    --to=matthias.fischer@ipfire.org \
    --cc=development@lists.ipfire.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox