Linux Archive

Linux Archive (http://www.linux-archive.org/)
-   Debian User (http://www.linux-archive.org/debian-user/)
-   -   buffer bloat or Squid issue ? (http://www.linux-archive.org/debian-user/678480-buffer-bloat-squid-issue.html)

Muhammad Yousuf Khan 06-29-2012 12:11 PM

buffer bloat or Squid issue ?
 
i am using squid 2.7 stable 9 with pdns as Caching DNS.
using KVM qemu VM as a proxyserver.

i have 4 Mbps internet link. when ever any client start download any
big file my squid gives the full threshold to the downloading session
and on other computers people see

The following error was encountered while trying to retrieve the URL:
http://www.google.com.pk/

Unable to determine IP address from host name www.google.com.pk

The DNS server returned:

Server Failure: The name server was unable to process this query.
This means that the cache was not able to resolve the hostname
presented in the URL. Check if the address is correct.

Your cache administrator is webmaster.

sorry to compare ISA here because i recently been shifted from ISA to
squid so i have to reference that in ISA if some one downloads then it
also handls other quires. of course bandwidth has been distributed and
performance was slow but didn't give the full threshold to just one
client.

when i ping to any outside host it gives me 4 figure value/

delay pool, i dont wana use it.

so i dont know how to track it down. please help me out.

Thanks,


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/CAGWVfMnGZssbk1CNX_v_66YBM6u5A98H1L=ibbXVA1KkB=f2c Q@mail.gmail.com

Muhammad Yousuf Khan 06-29-2012 12:11 PM

buffer bloat or Squid issue ?
 
see the ping delay of my proxy server.
see first few lines which is before download when i start download a
huge file it when to 3000 which is too much and this is the only
reason i think why it is happening. now at the bottem of this ping
responce you will see when i cancled the download job things back to
normal.

64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7361
ttl=54 time=70.3 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7368
ttl=54 time=68.0 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7369
ttl=54 time=70.6 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7370
ttl=54 time=196 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7371
ttl=54 time=787 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7372
ttl=54 time=1141 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7373
ttl=54 time=1689 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7374
ttl=54 time=2113 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7375
ttl=54 time=2262 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7376
ttl=54 time=2806 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7377
ttl=54 time=3083 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7378
ttl=54 time=3181 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7386
ttl=54 time=1846 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7387
ttl=54 time=1923 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7388
ttl=54 time=1807 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7389
ttl=54 time=831 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7390
ttl=54 time=71.2 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7391
ttl=54 time=69.2 ms
64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7392
ttl=54 time=69.3 ms


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: CAGWVfMkEx1ELOCUdtVNAnV4NnUbAbpOKJYF1kxeWGUbf0CJqS A@mail.gmail.com">http://lists.debian.org/CAGWVfMkEx1ELOCUdtVNAnV4NnUbAbpOKJYF1kxeWGUbf0CJqS A@mail.gmail.com

Muhammad Yousuf Khan 06-29-2012 01:15 PM

buffer bloat or Squid issue ?
 
i have done some test and here are some details and results

i am using two machine

1, Gateway IPcop (linux)
2. Debian lenny (squid)

i am using download manager to download a 50MB file.

IPCOP
-----------
when i do it VIA IPCOP my download burst rate up to 270 KB
not ping delay and other can also brows easily.

Squid on Lenny
------------------------

VIA SQuid (proxy mode) my download reach 365 which is full throughput
and faster then IPCOP but
ping delay reach 4000 which is considered almost near to death.

and no other users can browse and getting time out message on there browser.


i think this shows that issue is with squid box and i don't know
weather i have to tweak the squid or TCP buffer or anything

i am lost please help.

Thanks,

On Fri, Jun 29, 2012 at 5:11 PM, Muhammad Yousuf Khan <sirtcp@gmail.com> wrote:
> see the ping delay of my proxy server.
> see first few lines which is before download when i start download a
> huge file it when to 3000 which is too much and this is the only
> reason i think why it is happening. now at the bottem of this ping
> responce you will see when i cancled the download job things back to
> normal.
>
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7361
> ttl=54 time=70.3 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7368
> ttl=54 time=68.0 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7369
> ttl=54 time=70.6 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7370
> ttl=54 time=196 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7371
> ttl=54 time=787 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7372
> ttl=54 time=1141 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7373
> ttl=54 time=1689 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7374
> ttl=54 time=2113 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7375
> ttl=54 time=2262 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7376
> ttl=54 time=2806 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7377
> ttl=54 time=3083 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7378
> ttl=54 time=3181 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7386
> ttl=54 time=1846 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7387
> ttl=54 time=1923 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7388
> ttl=54 time=1807 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7389
> ttl=54 time=831 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7390
> ttl=54 time=71.2 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7391
> ttl=54 time=69.2 ms
> 64 bytes from del01s05-in-f7.1e100.net (74.125.236.7): icmp_req=7392
> ttl=54 time=69.3 ms


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/CAGWVfM=P10Kr1TtviRKXiNXm2SHSXsdgWN1W3i=kik7Tz+Ouy Q@mail.gmail.com

Tom Grace 06-29-2012 02:58 PM

buffer bloat or Squid issue ?
 
On 29/06/12 14:15, Muhammad Yousuf Khan wrote:
> i have done some test and here are some details and results
>
> i am using two machine
>
> 1, Gateway IPcop (linux)
> 2. Debian lenny (squid)
>
> i am using download manager to download a 50MB file.
>
> IPCOP
> -----------
> when i do it VIA IPCOP my download burst rate up to 270 KB
> not ping delay and other can also brows easily.
>
> Squid on Lenny
> ------------------------
>
> VIA SQuid (proxy mode) my download reach 365 which is full throughput
> and faster then IPCOP but
> ping delay reach 4000 which is considered almost near to death.
>
> and no other users can browse and getting time out message on there browser.
>
>
> i think this shows that issue is with squid box and i don't know
> weather i have to tweak the squid or TCP buffer or anything
You may want to throttle the connection in Squid. It seems like the
connection is saturated, though under normal circumstances it that
wouldn't cause the level of trouble that you are seeing. It might be
worth investigating the settings on the IPCOP box.
Your theory about buffers could well be correct, but probably the
simplest thing to do is limit the maximum speed to around 95% of the
available bandwidth (may require some experimentation).



--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 4FEDC297.7080302@deathbycomputers.co.uk">http://lists.debian.org/4FEDC297.7080302@deathbycomputers.co.uk

Muhammad Yousuf Khan 06-29-2012 03:18 PM

buffer bloat or Squid issue ?
 
On Fri, Jun 29, 2012 at 7:58 PM, Tom Grace
<lists-in@deathbycomputers.co.uk> wrote:
> On 29/06/12 14:15, Muhammad Yousuf Khan wrote:
>> i have done some test and here are some details and results
>>
>> *i am using two machine
>>
>> 1, Gateway IPcop (linux)
>> 2. Debian lenny (squid)
>>
>> i am using download manager to download a 50MB file.
>>
>> IPCOP
>> -----------
>> when i do it VIA IPCOP my download *burst rate up to 270 KB
>> not ping delay and other can also brows easily.
>>
>> Squid on Lenny
>> ------------------------
>>
>> VIA SQuid (proxy mode) my download reach 365 which is full throughput
>> and faster then IPCOP but
>> ping delay reach 4000 which is considered almost near to death.
>>
>> and no other users can browse and getting time out message on there browser.
>>
>>
>> i think this shows that issue is with squid box and i don't know
>> weather i have to tweak the squid or TCP buffer or anything
> You may want to throttle the connection in Squid. It seems like the
> connection is saturated, though under normal circumstances it that
> wouldn't cause the level of trouble that you are seeing. It might be
> worth investigating the settings on the IPCOP box.
> Your theory about buffers could well be correct, but probably the
> simplest thing to do is limit the maximum speed to around 95% of the
> available bandwidth (may require some experimentation).

for that i have to read 3 delaypools :(, means lots of testing..............

is there anyway or utility which can help me in this regard on OS level.

>
>
>
> --
> To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
> Archive: http://lists.debian.org/4FEDC297.7080302@deathbycomputers.co.uk
>


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: CAGWVfMmxQORG9Z6Py0m4Z9ASymVwR3sMx5xqEm8OjQLms8hSx w@mail.gmail.com">http://lists.debian.org/CAGWVfMmxQORG9Z6Py0m4Z9ASymVwR3sMx5xqEm8OjQLms8hSx w@mail.gmail.com

Muhammad Yousuf Khan 06-29-2012 03:21 PM

buffer bloat or Squid issue ?
 
FYI : IPCOP currently sets to masqurading. i am not using the squid
facility right now.

Thanks

On Fri, Jun 29, 2012 at 8:18 PM, Muhammad Yousuf Khan <sirtcp@gmail.com> wrote:
> On Fri, Jun 29, 2012 at 7:58 PM, Tom Grace
> <lists-in@deathbycomputers.co.uk> wrote:
>> On 29/06/12 14:15, Muhammad Yousuf Khan wrote:
>>> i have done some test and here are some details and results
>>>
>>> *i am using two machine
>>>
>>> 1, Gateway IPcop (linux)
>>> 2. Debian lenny (squid)
>>>
>>> i am using download manager to download a 50MB file.
>>>
>>> IPCOP
>>> -----------
>>> when i do it VIA IPCOP my download *burst rate up to 270 KB
>>> not ping delay and other can also brows easily.
>>>
>>> Squid on Lenny
>>> ------------------------
>>>
>>> VIA SQuid (proxy mode) my download reach 365 which is full throughput
>>> and faster then IPCOP but
>>> ping delay reach 4000 which is considered almost near to death.
>>>
>>> and no other users can browse and getting time out message on there browser.
>>>
>>>
>>> i think this shows that issue is with squid box and i don't know
>>> weather i have to tweak the squid or TCP buffer or anything
>> You may want to throttle the connection in Squid. It seems like the
>> connection is saturated, though under normal circumstances it that
>> wouldn't cause the level of trouble that you are seeing. It might be
>> worth investigating the settings on the IPCOP box.
>> Your theory about buffers could well be correct, but probably the
>> simplest thing to do is limit the maximum speed to around 95% of the
>> available bandwidth (may require some experimentation).
>
> for that i have to read 3 delaypools :(, means lots of testing..............
>
> is there anyway or utility which can help me in this regard on OS level.
>
>>
>>
>>
>> --
>> To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
>> with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
>> Archive: http://lists.debian.org/4FEDC297.7080302@deathbycomputers.co.uk
>>


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: CAGWVfMkjKQHhvZ9PZC_55kZ8qQ6P+BMvQnOC-+EQpGB_jx2CxQ@mail.gmail.com">http://lists.debian.org/CAGWVfMkjKQHhvZ9PZC_55kZ8qQ6P+BMvQnOC-+EQpGB_jx2CxQ@mail.gmail.com


All times are GMT. The time now is 11:35 AM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.