You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@httpd.apache.org by "David P. Donahue" <dd...@ccs.neu.edu> on 2004/09/15 17:01:14 UTC
[users@httpd] Narrowing Down A Strange Problem
Before I get into the details of my problem, let me specify that it is
my belief that the Apache software (version 1.3.31) is _not_ at fault
for this and I am just looking to either: 1) confirm that and look for
solutions elsewhere; or 2) be proven wrong and hopefully solve the
problem here and now.
The Scenario:
I have a simple website with a bunch of static files that is in place
for no other reason than to test connectivity and other issues before I
work on the new site(s). For the curious, this site is located at
http://www.cyber0ne.com and, keep in mind, a lot of the functionality of
the site itself is broken and it's a crappy old site from when I was in
college.
The Symptoms:
Browse to the site from the localhost (0 hops): no problems.
Browse to the site from LAN hosts (1 hop): no problems.
Browse to the site from hosts in the general metropolitan vecinity
(few hops): no problems.
Browse to the site from across the country (many hops): strange
problems seen.
Requests are sent to the server, noted in its logs, responded to as
usual. However, only responses (just simple GET requests for static
files here) with fewer than approximately 1250 bytes are received by the
client. Server logs show nothing out of the ordinary, but responses
greater than approximately 1250 bytes are never received by the client
and time out.
The Question(s):
Has anyone seen anything like this before?
Can you think of _anything_ in Apache which may result in such
behavior?
Can you, from wherever you are, access the site? (The root index page
exceeds the 1250-ish-byte limitation so, if you can see that, you should
be good.)
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
Re: [users@httpd] Narrowing Down A Strange Problem
Posted by Alexander Stoll <te...@gmx.de>.
suomi schrieb:
> a partial solution was that i set the MTU on all interfaces of our
> servers to 1468 (normally 1500), which was the maximum MTU supported
> by the backup network of the ISP.
It is a full solution if the lowest MTU/MRU of a link without proper
PMTU discovery is known. The only impact is reduced performance since
you need more packets for the same amount of transfered data...
> if you are in the same troubles: try to detect the maximum MTU of the
> ISP in question and set your servers to that value. remember
> however, that this is a partial solution ownly, because if one of
> your customers uploads a file to your server with MTUs larger than
> the maximum, he will fail all the same.
This is not true! When the interface MTU of the server is lowered
appropriate, all TCP traffic is save, because in the 3-way TCP-Handshake
the server never negotiates a higher MTU than the the lowest of both
partners, so even a client with usual MTU of 1500 can safely connect and
transfer any amount of data.
Things got stuck when on a transfer net a lower MTU is needed and the
border router is denied to send ICMP FN packets, mostly for a false
sense of security countermeasures.
Again: Providers with those braindead configurations (even non RFC
compliant) should pull the plug and being told so!
Re: [users@httpd] Narrowing Down A Strange Problem
Posted by suomi <ap...@ayni.com>.
Hi Dave
similar problem here last week:
from the internet, you can send a request to our web, but never get it
answered.
our problem: the primary line of our ISP had broken down and we went
automatically through the ISP's backup network. this backup network had
this in particular that it did not support maximum-MTU-dectection
according to rfc xyz.
a partial solution was that i set the MTU on all interfaces of our
servers to 1468 (normally 1500), which was the maximum MTU supported by
the backup network of the ISP.
if you are in the same troubles: try to detect the maximum MTU of the
ISP in question and set your servers to that value. remember however,
that this is a partial solution ownly, because if one of your customers
uploads a file to your server with MTUs larger than the maximum, he will
fail all the same.
suomi
David P. Donahue wrote:
> Before I get into the details of my problem, let me specify that it is
> my belief that the Apache software (version 1.3.31) is _not_ at fault
> for this and I am just looking to either: 1) confirm that and look for
> solutions elsewhere; or 2) be proven wrong and hopefully solve the
> problem here and now.
>
> The Scenario:
> I have a simple website with a bunch of static files that is in place
> for no other reason than to test connectivity and other issues before I
> work on the new site(s). For the curious, this site is located at
> http://www.cyber0ne.com and, keep in mind, a lot of the functionality of
> the site itself is broken and it's a crappy old site from when I was in
> college.
>
> The Symptoms:
> Browse to the site from the localhost (0 hops): no problems.
> Browse to the site from LAN hosts (1 hop): no problems.
> Browse to the site from hosts in the general metropolitan vecinity
> (few hops): no problems.
> Browse to the site from across the country (many hops): strange
> problems seen.
> Requests are sent to the server, noted in its logs, responded to as
> usual. However, only responses (just simple GET requests for static
> files here) with fewer than approximately 1250 bytes are received by the
> client. Server logs show nothing out of the ordinary, but responses
> greater than approximately 1250 bytes are never received by the client
> and time out.
>
> The Question(s):
> Has anyone seen anything like this before?
> Can you think of _anything_ in Apache which may result in such
> behavior?
> Can you, from wherever you are, access the site? (The root index page
> exceeds the 1250-ish-byte limitation so, if you can see that, you should
> be good.)
>
>
> ---------------------------------------------------------------------
> The official User-To-User support forum of the Apache HTTP Server Project.
> See <URL:http://httpd.apache.org/userslist.html> for more info.
> To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
> " from the digest: users-digest-unsubscribe@httpd.apache.org
> For additional commands, e-mail: users-help@httpd.apache.org
>
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
Re: [users@httpd] Narrowing Down A Strange Problem
Posted by Jim Maul <jm...@elih.org>.
Quoting "David P. Donahue" <dd...@ccs.neu.edu>:
> Before I get into the details of my problem, let me specify that it is
> my belief that the Apache software (version 1.3.31) is _not_ at fault
> for this and I am just looking to either: 1) confirm that and look for
> solutions elsewhere; or 2) be proven wrong and hopefully solve the
> problem here and now.
>
> The Scenario:
> I have a simple website with a bunch of static files that is in place
> for no other reason than to test connectivity and other issues before I
> work on the new site(s). For the curious, this site is located at
> http://www.cyber0ne.com and, keep in mind, a lot of the functionality of
> the site itself is broken and it's a crappy old site from when I was in
> college.
>
> The Symptoms:
> Browse to the site from the localhost (0 hops): no problems.
> Browse to the site from LAN hosts (1 hop): no problems.
> Browse to the site from hosts in the general metropolitan vecinity
> (few hops): no problems.
> Browse to the site from across the country (many hops): strange
> problems seen.
> Requests are sent to the server, noted in its logs, responded to as
> usual. However, only responses (just simple GET requests for static
> files here) with fewer than approximately 1250 bytes are received by the
> client. Server logs show nothing out of the ordinary, but responses
> greater than approximately 1250 bytes are never received by the client
> and time out.
>
> The Question(s):
> Has anyone seen anything like this before?
Maybe, but not me.
> Can you think of _anything_ in Apache which may result in such
> behavior?
If people closer to you can view the page fine, my guess would be something
wrong with an ISP further down the line.
> Can you, from wherever you are, access the site? (The root index page
> exceeds the 1250-ish-byte limitation so, if you can see that, you should
> be good.)
>
Im in NY, and no, i cant view the page.
-Jim
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
Re: [users@httpd] Narrowing Down A Strange Problem
Posted by Dave Floyd <da...@pa.press.net>.
> Can you, from wherever you are, access the site? (The root index page
>exceeds the 1250-ish-byte limitation so, if you can see that, you should
>be good.)
David,
There is no problem viewing the site from London. Here is the
traceroute:
Find route from: 144.178.105.208
to: www.cyber0ne.com (65.4.93.122), Max 30 hops, 40 byte packets
Host Names truncated to 32 bytes
1 144.178.105.3 (144.178.105.3 ): 7ms
0ms 0ms
2 frost.router.press.net. (144.178.106.15 ): 2ms
2ms 2ms
3 e0340053-a3-0-11.ctf-18.ctf.lon. (212.161.49.177 ): 164ms
218ms 96ms
4 g2-0.cr3.lon.ctf.lon.uk.colt.net (195.110.65.1 ): 20ms
30ms 18ms
5 pos2-0-super-grover.lon.router.c (212.74.64.29 ): 23ms
16ms 4ms
6 pos1-0-kenny.nyc.router.colt.net (212.74.64.174 ): 73ms
73ms 73ms
7 ge-8-2-310.ipcolo2.newyork1.leve (63.214.53.253 ): 73ms
85ms 106ms
8 ge-7-0-0.bbr2.newyork1.level3.ne (4.68.115.137 ): 131ms
101ms 152ms
9 ge-7-0-0.gar4.newyork1.level3.ne (64.159.4.150 ): 159ms
171ms 112ms
10 uunet-level3-oc48.newyork1.level (209.244.160.182): 118ms
73ms 83ms
11 0.so-6-0-0.xl1.nyc4.alter.net. (152.63.21.78 ): 73ms
75ms 73ms
12 0.so-4-0-0.tl1.nyc9.alter.net. (152.63.0.173 ): 73ms
78ms 86ms
13 0.so-4-1-0.tl1.atl5.alter.net. (152.63.10.129 ): 91ms
91ms 91ms
14 0.so-6-0-0.xl1.atl5.alter.net. (152.63.10.101 ): 91ms
91ms 91ms
15 0.so-6-0-0.gw13.atl5.alter.net. (152.63.84.105 ): 117ms
91ms 101ms
16 bellsouth-atl5-gw.customer.alter (157.130.71.170 ): 92ms
91ms 112ms
17 axr00asm-1-0-0.bellsouth.net. (65.83.236.3 ): 94ms
107ms 95ms
18 ixc00cae-8-1-1.bellsouth.net. (65.83.237.53 ): 108ms
99ms 99ms
19 205.152.135.81 (205.152.135.81 ): 117ms
106ms 100ms
20 68.152.231.138 (68.152.231.138 ): 100ms
110ms 115ms
21 * * *
22 * * *
23 * * *
24 * * *
25 * * *
26 * * *
27 * * *
28 * * *
29 * * *
30 * * *
*Maximum Hop Count Exceeded!!*
I see a couple of the other traceroutes supplied also timed
out after the same router. This presumably means the router is just
configured to reject traces, but it lets www connections through. You
may have a problem with a blackhole configuration on one of the
routers in a particular chain?
HTH
Dave
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
RE: [users@httpd] Narrowing Down A Strange Problem
Posted by "David P. Donahue" <dd...@ccs.neu.edu>.
I've seen similar behavior over telnet to port 80. A small sample of
results is below:
GET /index.html
Hangs until cancelled. Note, index.html is greater than 1250 bytes.
GET /what/
No problems. Directory listing is fewer than 1250 bytes.
Get /what/pictures/
No problems. Directory listing is fewer than 1250 bytes.
Get /what/media/
Hangs until cancelled. Directory listing is greater than 1250 bytes.
And so on. You'll also notice that the response to the invalid command
has fewer than 1250 bytes.
>
> in a browser i get no response..it sits for a minute or two
> trying to load the page and then just times out. I tried to
> telnet to port 80 also..
>
> [root@external qmail-smtpd]# telnet www.cyber0ne.com 80
> Trying 65.4.93.122...
> Connected to www.cyber0ne.com.
> Escape character is '^]'.
> GET
>
> Connection closed by foreign host.
>
>
> Note the "Connection closed" message only came up after i hit
> enter for
> a second
> time. After the GET <CR> it just hung for a while..
>
> Also note that if you send an invalid command, ie:
>
> [root@external qmail-smtpd]# telnet www.cyber0ne.com 80
> Trying 65.4.93.122...
> Connected to www.cyber0ne.com.
> Escape character is '^]'.
> asdf
> <!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN">
> <HTML><HEAD>
> <TITLE>403 Forbidden</TITLE>
> </HEAD><BODY>
> <H1>Forbidden</H1>
> You don't have permission to access /
> on this server.<P>
> <HR>
> <ADDRESS>Apache/1.3.31 Server at www.cyber0ne.com Port
> 80</ADDRESS> </BODY></HTML> Connection closed by foreign host.
>
> The server does respond right away with an error message so
> its definitely working for atleast some requests.
>
> -Jim
>
>
>
> ---------------------------------------------------------------------
> The official User-To-User support forum of the Apache HTTP
> Server Project. See
> <URL:http://httpd.apache.org/userslist.html> for more info.
> To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
> " from the digest: users-digest-unsubscribe@httpd.apache.org
> For additional commands, e-mail: users-help@httpd.apache.org
>
>
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
RE: [users@httpd] Narrowing Down A Strange Problem
Posted by Jim Maul <jm...@elih.org>.
Quoting "David P. Donahue" <dd...@ccs.neu.edu>:
> So far people who can view the site are in:
> South Carolina
> North Carolina
> Missouri
> Arkansas
> And people who can not are int:
> Massachusetts
> New York
> California
>
> The only pattern I can derive from that data is the number of hops in
> the route.
>
>
in a browser i get no response..it sits for a minute or two trying to load the
page and then just times out. I tried to telnet to port 80 also..
[root@external qmail-smtpd]# telnet www.cyber0ne.com 80
Trying 65.4.93.122...
Connected to www.cyber0ne.com.
Escape character is '^]'.
GET
Connection closed by foreign host.
Note the "Connection closed" message only came up after i hit enter for
a second
time. After the GET <CR> it just hung for a while..
Also note that if you send an invalid command, ie:
[root@external qmail-smtpd]# telnet www.cyber0ne.com 80
Trying 65.4.93.122...
Connected to www.cyber0ne.com.
Escape character is '^]'.
asdf
<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN">
<HTML><HEAD>
<TITLE>403 Forbidden</TITLE>
</HEAD><BODY>
<H1>Forbidden</H1>
You don't have permission to access /
on this server.<P>
<HR>
<ADDRESS>Apache/1.3.31 Server at www.cyber0ne.com Port 80</ADDRESS>
</BODY></HTML>
Connection closed by foreign host.
The server does respond right away with an error message so its definitely
working for atleast some requests.
-Jim
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
RE: [users@httpd] Narrowing Down A Strange Problem
Posted by "David P. Donahue" <dd...@ccs.neu.edu>.
So far people who can view the site are in:
South Carolina
North Carolina
Missouri
Arkansas
And people who can not are int:
Massachusetts
New York
California
The only pattern I can derive from that data is the number of hops in
the route.
-----Original Message-----
From: Aman Raheja [mailto:araheja@techquotes.com]
Sent: Wednesday, September 15, 2004 11:53 AM
To: users@httpd.apache.org
Subject: Re: [users@httpd] Narrowing Down A Strange Problem
From how many different locations / ISPs / people you have seen this
problem?
I am from St Louis and am able to see the site.
Thanks
David P. Donahue wrote:
>>>Did work fine for me. What is the error people get
>>>who can't access it for whatever reason and how do
>>>you know that there is this 1250 byte limit?
>>>Aman Raheja
>>>
>>>
>
>No error is received for >1250b responses, it just never gets to the
>client. I haven't narrowed it down to a specific byte size, but the
>pattern of positive results vs. time-outs seems to hold at
>approximately 1250 bytes. I've made a large number of GET requests and
>the pattern has held. Even directory listings from Apache itself fit
>the pattern if there are enough files in the directory to bring the
>generated response above the 1250-ish-byte line.
>
>If you were able to view the site, might I ask your general location
>and/or ISP?
>
>
>David P. Donahue wrote:
>
>
>
>>Before I get into the details of my problem, let me specify that it is
>>my belief that the Apache software (version 1.3.31) is _not_ at fault
>>for this and I am just looking to either: 1) confirm that and look for
>>solutions elsewhere; or 2) be proven wrong and hopefully solve the
>>problem here and now.
>>
>>The Scenario:
>> I have a simple website with a bunch of static files that is in place
>>
>>
>
>
>
>>for no other reason than to test connectivity and other issues before
>>I
>>
>>
>
>
>
>>work on the new site(s). For the curious, this site is located at
>>http://www.cyber0ne.com and, keep in mind, a lot of the functionality
>>of the site itself is broken and it's a crappy old site from when I
was
>>
>>
>
>
>
>>in college.
>>
>>The Symptoms:
>> Browse to the site from the localhost (0 hops): no problems. Browse
>>to the site from LAN hosts (1 hop): no problems. Browse to the site
>>from hosts in the general metropolitan vecinity (few hops): no
>>problems. Browse to the site from across the country (many hops):
>>strange problems seen.
>> Requests are sent to the server, noted in its logs, responded to as
>>
>>
>
>
>
>>usual. However, only responses (just simple GET requests for static
>>files here) with fewer than approximately 1250 bytes are received by
>>the client. Server logs show nothing out of the ordinary, but
>>responses greater than approximately 1250 bytes are never received by
>>the client and time out.
>>
>>The Question(s):
>> Has anyone seen anything like this before?
>> Can you think of _anything_ in Apache which may result in such
>>behavior?
>> Can you, from wherever you are, access the site? (The root index page
>>
>>
>
>
>
>>exceeds the 1250-ish-byte limitation so, if you can see that, you
>>should be good.)
>>
>>
>>
>>
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server
Project. See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
Re: [users@httpd] Narrowing Down A Strange Problem
Posted by Aman Raheja <ar...@techquotes.com>.
From how many different locations / ISPs / people you have seen this
problem?
I am from St Louis and am able to see the site.
Thanks
David P. Donahue wrote:
>>>Did work fine for me. What is the error people get
>>>who can't access it for whatever reason and how do
>>>you know that there is this 1250 byte limit?
>>>Aman Raheja
>>>
>>>
>
>No error is received for >1250b responses, it just never gets to the
>client.
>I haven't narrowed it down to a specific byte size, but the pattern of
>positive results vs. time-outs seems to hold at approximately 1250
>bytes. I've made a large number of GET requests and the pattern has
>held. Even directory listings from Apache itself fit the pattern if
>there are enough files in the directory to bring the generated response
>above the 1250-ish-byte line.
>
>If you were able to view the site, might I ask your general location
>and/or ISP?
>
>
>David P. Donahue wrote:
>
>
>
>>Before I get into the details of my problem, let me specify that it is
>>my belief that the Apache software (version 1.3.31) is _not_ at fault
>>for this and I am just looking to either: 1) confirm that and look for
>>solutions elsewhere; or 2) be proven wrong and hopefully solve the
>>problem here and now.
>>
>>The Scenario:
>> I have a simple website with a bunch of static files that is in place
>>
>>
>
>
>
>>for no other reason than to test connectivity and other issues before I
>>
>>
>
>
>
>>work on the new site(s). For the curious, this site is located at
>>http://www.cyber0ne.com and, keep in mind, a lot of the functionality
>>of the site itself is broken and it's a crappy old site from when I was
>>
>>
>
>
>
>>in college.
>>
>>The Symptoms:
>> Browse to the site from the localhost (0 hops): no problems.
>> Browse to the site from LAN hosts (1 hop): no problems.
>> Browse to the site from hosts in the general metropolitan vecinity
>>(few hops): no problems.
>> Browse to the site from across the country (many hops): strange
>>problems seen.
>> Requests are sent to the server, noted in its logs, responded to as
>>
>>
>
>
>
>>usual. However, only responses (just simple GET requests for static
>>files here) with fewer than approximately 1250 bytes are received by
>>the client. Server logs show nothing out of the ordinary, but
>>responses greater than approximately 1250 bytes are never received by
>>the client and time out.
>>
>>The Question(s):
>> Has anyone seen anything like this before?
>> Can you think of _anything_ in Apache which may result in such
>>behavior?
>> Can you, from wherever you are, access the site? (The root index page
>>
>>
>
>
>
>>exceeds the 1250-ish-byte limitation so, if you can see that, you
>>should be good.)
>>
>>
>>
>>
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
Re: [users@httpd] Narrowing Down A Strange Problem
Posted by Alexander Stoll <te...@gmx.de>.
Jim Maul schrieb:
> I would just like to add that this should only have to be done on some dsl
> lines. We had a similar problem on a dsl line that we have where
> outlook would
> refuse to send out an email if it was over a certain size (the mtu
> setting). We
> had to go around to all workstations and change the mtu from 1500 to
> 1492. What
> a pita.
hmm, not so easy... these days it is prefered to set the DF (don´t
fragment bit) to keep the load on the routers as low as possible... but
ANY link on the route that has a lower mtu and prohibits ICMP FN packets
causes a "blackhole" for transit packets because the TCP sessions stuck.
The problem is braindead configuration of backbone providers...
Re: [users@httpd] Narrowing Down A Strange Problem
Posted by Jeff White <jl...@earthlink.net>.
From: "Jim Maul"
>> I am interested in your suggestion
>> to "lower the MTU of your server."
> We had to go around to all workstations
> and change the mtu from 1500
> to 1492. What a pita.
I do not know what type of network
one has, but if it's Windows, perhaps
one could have done it from one
workstation or via the network logon
scripts.
Configuring the MTU for all Network Adapters
http://www.microsoft.com/technet/community/scriptcenter/network/scnet128.mspx
WMI Scripting Primer
(From the online Microsoft Windows 2000 Scripting Guide)
http://www.microsoft.com/resources/documentation/windows/2000/server/scriptguide/en-us/sas_wmi_overview.mspx
Google WMI Logon
http://groups.google.com/groups?as_q=WMI%20logon&safe=images&ie=UTF-8&as_ugroup=microsoft.public.scripting.vbscript&lr=&hl=en
Jeff
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
Re: [users@httpd] Narrowing Down A Strange Problem
Posted by Jim Maul <jm...@elih.org>.
Quoting Alexander Stoll <te...@gmx.de>:
> David P. Donahue schrieb:
>
>> I am interested in your suggestion to "lower the MTU of your server."
>> This is new to me and I would love to add it to my testing. How would I
>> lower this value? Note, I am running Apache 1.3.31 on Slackware 10.0.
>
> OK, here we go...
> This is something your OS has to provide and is independend from Apache...
> I don´t know slackware distribution but basic tools must be present...
>
> if you become root on the system and fire up a "ifconfig" something like
>
> eth0 Link encap:Ethernet HWaddr 00:D0:B7:06:2B:4D
> inet addr:10.0.0.130 Bcast:10.0.0.255 Mask:255.255.255.0
> UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
>
> should appear, if you prefer ip, type in "ip link show" an something like
>
> 1: lo: <LOOPBACK,UP> mtu 16436 qdisc noqueue
> link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> 2: eth0: <BROADCAST,MULTICAST,ALLMULTI,UP> mtu 1500 qdisc pfifo_fast
> qlen 1000
> link/ether 00:d0:b7:06:2b:4d brd ff:ff:ff:ff:ff:ff
>
> this indicates a mtu of 1500 (normal ethernet default)
>
> If you haven´t access to the console of the server, be aware of that any
> changes who interrupt network connectivity could cause no more access to
> your server! So be careful... ;-)
>
> with present values
> "ifconfig eth0 down; ifconfig eth0 10.0.0.130 netmask 255.255.255.0
> broadcast 10.0.0.255 mtu 1250 up" (in one line!!!)
> brings up the interface with desired mtu
>
> alternative with ip:
>
> "ip link set eth0 mtu 1250"
>
> This should be prefered if ip is present on your server since this
> should be non destructive to your network connectivity. Replace values
> for address, netmask, broadcast and interface name with your local needs...
>
> REMEMBER! Don´t hammer on my mailbox when a mistyped command destroys
> network access to your server until rebooted... ;-)))
>
I would just like to add that this should only have to be done on some dsl
lines. We had a similar problem on a dsl line that we have where
outlook would
refuse to send out an email if it was over a certain size (the mtu
setting). We
had to go around to all workstations and change the mtu from 1500 to
1492. What
a pita.
-Jim
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
Re: [users@httpd] Narrowing Down A Strange Problem
Posted by Alexander Stoll <te...@gmx.de>.
David P. Donahue schrieb:
> Well that didn't take long...
>
> Ok, I got anxious and tried it remotely, using /sbin/ip. And what do
> you know, it _worked_! Further testing is required to confirm this
> enough that I'm comfortable with it, but if I could hug you right now I
> would.
>
> I'll still stay on track with my ISP to see if the problem itself can be
> fixed, but this seems to be an excellent work-around for my more
> immediate needs. Thank you!
>
>
> Regards,
> David P. Donahue
You´re welcome, if you could track down the bad hop that causes this
mess, you could set a dedicated route with lower mtu for that, better
force the peering partners of the bad link to to pull the plug for that...
Since it turned out to be a fully network related issue, feel free to
continue conversation via PM and not bother the rest of the list with OT
stuff...
regards, AS
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
RE: [users@httpd] Narrowing Down A Strange Problem
Posted by "David P. Donahue" <dd...@ccs.neu.edu>.
Well that didn't take long...
Ok, I got anxious and tried it remotely, using /sbin/ip. And what do
you know, it _worked_! Further testing is required to confirm this
enough that I'm comfortable with it, but if I could hug you right now I
would.
I'll still stay on track with my ISP to see if the problem itself can be
fixed, but this seems to be an excellent work-around for my more
immediate needs. Thank you!
Regards,
David P. Donahue
> -----Original Message-----
> From: David P. Donahue [mailto:ddonahue@ccs.neu.edu]
> Sent: Wednesday, September 15, 2004 2:08 PM
> To: users@httpd.apache.org
> Subject: RE: [users@httpd] Narrowing Down A Strange Problem
>
>
> I think I'll wait until I get home tonight (and have console
> access) to test this :)
>
> But you've taught me something new, thank you!
>
>
>
> > -----Original Message-----
> > From: Alexander Stoll [mailto:technoworx@gmx.de]
> > Sent: Wednesday, September 15, 2004 2:09 PM
> > To: users@httpd.apache.org
> > Subject: Re: [users@httpd] Narrowing Down A Strange Problem
> >
> >
> > David P. Donahue schrieb:
> >
> > > I am interested in your suggestion to "lower the MTU of
> > your server."
> > > This is new to me and I would love to add it to my testing.
> > How would
> > > I lower this value? Note, I am running Apache 1.3.31 on Slackware
> > > 10.0.
> >
> > OK, here we go...
> > This is something your OS has to provide and is independend
> > from Apache... I don´t know slackware distribution but basic
> > tools must be present...
> >
> > if you become root on the system and fire up a "ifconfig"
> > something like
> >
> > eth0 Link encap:Ethernet HWaddr 00:D0:B7:06:2B:4D
> > inet addr:10.0.0.130 Bcast:10.0.0.255
> Mask:255.255.255.0
> > UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
> >
> > should appear, if you prefer ip, type in "ip link show" an
> > something like
> >
> > 1: lo: <LOOPBACK,UP> mtu 16436 qdisc noqueue
> > link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> > 2: eth0: <BROADCAST,MULTICAST,ALLMULTI,UP> mtu 1500 qdisc pfifo_fast
> > qlen 1000
> > link/ether 00:d0:b7:06:2b:4d brd ff:ff:ff:ff:ff:ff
> >
> > this indicates a mtu of 1500 (normal ethernet default)
> >
> > If you haven´t access to the console of the server, be aware
> > of that any
> > changes who interrupt network connectivity could cause no
> > more access to
> > your server! So be careful... ;-)
> >
> > with present values
> > "ifconfig eth0 down; ifconfig eth0 10.0.0.130 netmask
> > 255.255.255.0 broadcast 10.0.0.255 mtu 1250 up" (in one
> > line!!!) brings up the interface with desired mtu
> >
> > alternative with ip:
> >
> > "ip link set eth0 mtu 1250"
> >
> > This should be prefered if ip is present on your server since this
> > should be non destructive to your network connectivity.
> > Replace values
> > for address, netmask, broadcast and interface name with your
> > local needs...
> >
> > REMEMBER! Don´t hammer on my mailbox when a mistyped
> command destroys
> > network access to your server until rebooted... ;-)))
> >
> > Best wishes, AS
> >
> >
> >
> ---------------------------------------------------------------------
> > The official User-To-User support forum of the Apache HTTP
> > Server Project. See
> > <URL:http://httpd.apache.org/userslist.html> for more info.
> > To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
> > " from the digest: users-digest-unsubscribe@httpd.apache.org
> > For additional commands, e-mail: users-help@httpd.apache.org
> >
> >
>
>
> ---------------------------------------------------------------------
> The official User-To-User support forum of the Apache HTTP
> Server Project. See
> <URL:http://httpd.apache.org/userslist.html> for more info.
> To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
> " from the digest: users-digest-unsubscribe@httpd.apache.org
> For additional commands, e-mail: users-help@httpd.apache.org
>
>
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
RE: [users@httpd] Narrowing Down A Strange Problem
Posted by "David P. Donahue" <dd...@ccs.neu.edu>.
I think I'll wait until I get home tonight (and have console access) to
test this :)
But you've taught me something new, thank you!
> -----Original Message-----
> From: Alexander Stoll [mailto:technoworx@gmx.de]
> Sent: Wednesday, September 15, 2004 2:09 PM
> To: users@httpd.apache.org
> Subject: Re: [users@httpd] Narrowing Down A Strange Problem
>
>
> David P. Donahue schrieb:
>
> > I am interested in your suggestion to "lower the MTU of
> your server."
> > This is new to me and I would love to add it to my testing.
> How would
> > I lower this value? Note, I am running Apache 1.3.31 on Slackware
> > 10.0.
>
> OK, here we go...
> This is something your OS has to provide and is independend
> from Apache... I don´t know slackware distribution but basic
> tools must be present...
>
> if you become root on the system and fire up a "ifconfig"
> something like
>
> eth0 Link encap:Ethernet HWaddr 00:D0:B7:06:2B:4D
> inet addr:10.0.0.130 Bcast:10.0.0.255 Mask:255.255.255.0
> UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
>
> should appear, if you prefer ip, type in "ip link show" an
> something like
>
> 1: lo: <LOOPBACK,UP> mtu 16436 qdisc noqueue
> link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> 2: eth0: <BROADCAST,MULTICAST,ALLMULTI,UP> mtu 1500 qdisc pfifo_fast
> qlen 1000
> link/ether 00:d0:b7:06:2b:4d brd ff:ff:ff:ff:ff:ff
>
> this indicates a mtu of 1500 (normal ethernet default)
>
> If you haven´t access to the console of the server, be aware
> of that any
> changes who interrupt network connectivity could cause no
> more access to
> your server! So be careful... ;-)
>
> with present values
> "ifconfig eth0 down; ifconfig eth0 10.0.0.130 netmask
> 255.255.255.0 broadcast 10.0.0.255 mtu 1250 up" (in one
> line!!!) brings up the interface with desired mtu
>
> alternative with ip:
>
> "ip link set eth0 mtu 1250"
>
> This should be prefered if ip is present on your server since this
> should be non destructive to your network connectivity.
> Replace values
> for address, netmask, broadcast and interface name with your
> local needs...
>
> REMEMBER! Don´t hammer on my mailbox when a mistyped command destroys
> network access to your server until rebooted... ;-)))
>
> Best wishes, AS
>
>
> ---------------------------------------------------------------------
> The official User-To-User support forum of the Apache HTTP
> Server Project. See
> <URL:http://httpd.apache.org/userslist.html> for more info.
> To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
> " from the digest: users-digest-unsubscribe@httpd.apache.org
> For additional commands, e-mail: users-help@httpd.apache.org
>
>
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
Re: [users@httpd] Narrowing Down A Strange Problem
Posted by Alexander Stoll <te...@gmx.de>.
David P. Donahue schrieb:
> I am interested in your suggestion to "lower the MTU of your server."
> This is new to me and I would love to add it to my testing. How would I
> lower this value? Note, I am running Apache 1.3.31 on Slackware 10.0.
OK, here we go...
This is something your OS has to provide and is independend from Apache...
I don´t know slackware distribution but basic tools must be present...
if you become root on the system and fire up a "ifconfig" something like
eth0 Link encap:Ethernet HWaddr 00:D0:B7:06:2B:4D
inet addr:10.0.0.130 Bcast:10.0.0.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
should appear, if you prefer ip, type in "ip link show" an something like
1: lo: <LOOPBACK,UP> mtu 16436 qdisc noqueue
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: eth0: <BROADCAST,MULTICAST,ALLMULTI,UP> mtu 1500 qdisc pfifo_fast
qlen 1000
link/ether 00:d0:b7:06:2b:4d brd ff:ff:ff:ff:ff:ff
this indicates a mtu of 1500 (normal ethernet default)
If you haven´t access to the console of the server, be aware of that any
changes who interrupt network connectivity could cause no more access to
your server! So be careful... ;-)
with present values
"ifconfig eth0 down; ifconfig eth0 10.0.0.130 netmask 255.255.255.0
broadcast 10.0.0.255 mtu 1250 up" (in one line!!!)
brings up the interface with desired mtu
alternative with ip:
"ip link set eth0 mtu 1250"
This should be prefered if ip is present on your server since this
should be non destructive to your network connectivity. Replace values
for address, netmask, broadcast and interface name with your local needs...
REMEMBER! Don´t hammer on my mailbox when a mistyped command destroys
network access to your server until rebooted... ;-)))
Best wishes, AS
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
RE: [users@httpd] Narrowing Down A Strange Problem
Posted by "David P. Donahue" <dd...@ccs.neu.edu>.
I am interested in your suggestion to "lower the MTU of your server."
This is new to me and I would love to add it to my testing. How would I
lower this value? Note, I am running Apache 1.3.31 on Slackware 10.0.
> Hi, this seem to be a MTU related problem on the network layer.
> Locations where access fails, should check with a full packet
> dump of a
> failed request. For some locations the packets go over a link
> with MTU
> of 1250b and proper PMTU discovery fails, so the TCP session simply
> stucks until timeout...
> Maybe you can determine the hop by comparing the traces...
> There not really much you can do with different impact:
> Generally lower the MTU of your server to 1250b which impacts the
> performance - more packets needed for same payload - or if
> you are able
> to determine exactly what ranges fail, intercept sessions with these
> sources and force a lower mtu for only these links on your server or
> router...
> If there is a braindead ISP on the hop who denies forwarding of "ICMP
> fragmentation needed" packets, contact him and tell him to
> pull the plug
> and never reconnect until he employed someone who has knowledge of
> TCP/IP networks...
>
>
> ---------------------------------------------------------------------
> The official User-To-User support forum of the Apache HTTP
> Server Project. See
> <URL:http://httpd.apache.org/userslist.html> for more info.
> To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
> " from the digest: users-digest-unsubscribe@httpd.apache.org
> For additional commands, e-mail: users-help@httpd.apache.org
>
>
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
Re: [users@httpd] Narrowing Down A Strange Problem
Posted by Alexander Stoll <te...@gmx.de>.
David P. Donahue schrieb:
> No error is received for >1250b responses, it just never gets to the
> client.
> I haven't narrowed it down to a specific byte size, but the pattern of
> positive results vs. time-outs seems to hold at approximately 1250
> bytes. I've made a large number of GET requests and the pattern has
> held. Even directory listings from Apache itself fit the pattern if
> there are enough files in the directory to bring the generated response
> above the 1250-ish-byte line.
>
> If you were able to view the site, might I ask your general location
> and/or ISP?
Hi, this seem to be a MTU related problem on the network layer.
Locations where access fails, should check with a full packet dump of a
failed request. For some locations the packets go over a link with MTU
of 1250b and proper PMTU discovery fails, so the TCP session simply
stucks until timeout...
Maybe you can determine the hop by comparing the traces...
There not really much you can do with different impact:
Generally lower the MTU of your server to 1250b which impacts the
performance - more packets needed for same payload - or if you are able
to determine exactly what ranges fail, intercept sessions with these
sources and force a lower mtu for only these links on your server or
router...
If there is a braindead ISP on the hop who denies forwarding of "ICMP
fragmentation needed" packets, contact him and tell him to pull the plug
and never reconnect until he employed someone who has knowledge of
TCP/IP networks...
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
RE: [users@httpd] Narrowing Down A Strange Problem
Posted by "David P. Donahue" <dd...@ccs.neu.edu>.
>>Did work fine for me. What is the error people get
>>who can't access it for whatever reason and how do
>>you know that there is this 1250 byte limit?
>>Aman Raheja
No error is received for >1250b responses, it just never gets to the
client.
I haven't narrowed it down to a specific byte size, but the pattern of
positive results vs. time-outs seems to hold at approximately 1250
bytes. I've made a large number of GET requests and the pattern has
held. Even directory listings from Apache itself fit the pattern if
there are enough files in the directory to bring the generated response
above the 1250-ish-byte line.
If you were able to view the site, might I ask your general location
and/or ISP?
David P. Donahue wrote:
>Before I get into the details of my problem, let me specify that it is
>my belief that the Apache software (version 1.3.31) is _not_ at fault
>for this and I am just looking to either: 1) confirm that and look for
>solutions elsewhere; or 2) be proven wrong and hopefully solve the
>problem here and now.
>
>The Scenario:
> I have a simple website with a bunch of static files that is in place
>for no other reason than to test connectivity and other issues before I
>work on the new site(s). For the curious, this site is located at
>http://www.cyber0ne.com and, keep in mind, a lot of the functionality
>of the site itself is broken and it's a crappy old site from when I was
>in college.
>
>The Symptoms:
> Browse to the site from the localhost (0 hops): no problems.
> Browse to the site from LAN hosts (1 hop): no problems.
> Browse to the site from hosts in the general metropolitan vecinity
>(few hops): no problems.
> Browse to the site from across the country (many hops): strange
>problems seen.
> Requests are sent to the server, noted in its logs, responded to as
>usual. However, only responses (just simple GET requests for static
>files here) with fewer than approximately 1250 bytes are received by
>the client. Server logs show nothing out of the ordinary, but
>responses greater than approximately 1250 bytes are never received by
>the client and time out.
>
>The Question(s):
> Has anyone seen anything like this before?
> Can you think of _anything_ in Apache which may result in such
>behavior?
> Can you, from wherever you are, access the site? (The root index page
>exceeds the 1250-ish-byte limitation so, if you can see that, you
>should be good.)
>
>
>---------------------------------------------------------------------
>The official User-To-User support forum of the Apache HTTP Server
>Project. See <URL:http://httpd.apache.org/userslist.html> for more
>info. To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
> " from the digest: users-digest-unsubscribe@httpd.apache.org
>For additional commands, e-mail: users-help@httpd.apache.org
>
>
>
>
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server
Project. See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org
Re: [users@httpd] Narrowing Down A Strange Problem
Posted by Aman Raheja <ar...@techquotes.com>.
Did work fine for me. What is the error people get who can't access it
for whatever reason and how do you know that there is this 1250 byte limit?
Aman Raheja
David P. Donahue wrote:
>Before I get into the details of my problem, let me specify that it is
>my belief that the Apache software (version 1.3.31) is _not_ at fault
>for this and I am just looking to either: 1) confirm that and look for
>solutions elsewhere; or 2) be proven wrong and hopefully solve the
>problem here and now.
>
>The Scenario:
> I have a simple website with a bunch of static files that is in place
>for no other reason than to test connectivity and other issues before I
>work on the new site(s). For the curious, this site is located at
>http://www.cyber0ne.com and, keep in mind, a lot of the functionality of
>the site itself is broken and it's a crappy old site from when I was in
>college.
>
>The Symptoms:
> Browse to the site from the localhost (0 hops): no problems.
> Browse to the site from LAN hosts (1 hop): no problems.
> Browse to the site from hosts in the general metropolitan vecinity
>(few hops): no problems.
> Browse to the site from across the country (many hops): strange
>problems seen.
> Requests are sent to the server, noted in its logs, responded to as
>usual. However, only responses (just simple GET requests for static
>files here) with fewer than approximately 1250 bytes are received by the
>client. Server logs show nothing out of the ordinary, but responses
>greater than approximately 1250 bytes are never received by the client
>and time out.
>
>The Question(s):
> Has anyone seen anything like this before?
> Can you think of _anything_ in Apache which may result in such
>behavior?
> Can you, from wherever you are, access the site? (The root index page
>exceeds the 1250-ish-byte limitation so, if you can see that, you should
>be good.)
>
>
>---------------------------------------------------------------------
>The official User-To-User support forum of the Apache HTTP Server Project.
>See <URL:http://httpd.apache.org/userslist.html> for more info.
>To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
> " from the digest: users-digest-unsubscribe@httpd.apache.org
>For additional commands, e-mail: users-help@httpd.apache.org
>
>
>
>
---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
" from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org