FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Debian > Debian User

 
 
LinkBack Thread Tools
 
Old 03-21-2012, 12:21 PM
Bonno Bloksma
 
Default packet loss after Lenny Squeeze upgrade

Hi,

I have a Debian server (hardware HP DL360) which I use for internal routing and which uses NAT to isolate a part of my network.
This machine has been routing for the past few years without any problems but lately there is a lot of packet loss for the traffic going to/through that server.

What has changed in Debian that could cause this and what can I tweak to get things going again?
What do I need to look at?

The server hardware does not seem to be the limiting factor, cpu is a max 10%, memory seems to be ok.

Bonno Bloksma


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 89D1798A7351D040B4E74E0A043C69D701079D65@HGLEXCH-01.tio.nl">http://lists.debian.org/89D1798A7351D040B4E74E0A043C69D701079D65@HGLEXCH-01.tio.nl
 
Old 03-21-2012, 02:23 PM
Frank
 
Default packet loss after Lenny Squeeze upgrade

Have you verified where the loss happens? How?

Am 21.03.2012 um 14:21 schrieb Bonno Bloksma <b.bloksma@tio.nl>:

> Hi,
>
> I have a Debian server (hardware HP DL360) which I use for internal routing and which uses NAT to isolate a part of my network.
> This machine has been routing for the past few years without any problems but lately there is a lot of packet loss for the traffic going to/through that server.
>
> What has changed in Debian that could cause this and what can I tweak to get things going again?
> What do I need to look at?
>
> The server hardware does not seem to be the limiting factor, cpu is a max 10%, memory seems to be ok.
>
> Bonno Bloksma
>
>
> --
> 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/89D1798A7351D040B4E74E0A043C69D701079D65@HGLEXCH-01.tio.nl
>


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: C56DE2C8-600E-490C-87B8-DCB362D86492@dead-link.org">http://lists.debian.org/C56DE2C8-600E-490C-87B8-DCB362D86492@dead-link.org
 
Old 03-21-2012, 03:00 PM
Bonno Bloksma
 
Default packet loss after Lenny Squeeze upgrade

Hello Frank,

Am 21.03.2012 um 14:21 schrieb Bonno Bloksma <b.bloksma@tio.nl>:

>> I have a Debian server (hardware HP DL360) which I use for internal routing and which uses NAT to isolate a part of my network.
>> This machine has been routing for the past few years without any problems but lately there is a lot of packet loss for the traffic going to/through that server.
>>
>> What has changed in Debian that could cause this and what can I tweak to get things going again?
>> What do I need to look at?
>>
>> The server hardware does not seem to be the limiting factor, cpu is a max 10%, memory seems to be ok.

> Have you verified where the loss happens? How?

What do you mean with WHERE the loss happens? I know what do to at the "normal" level where I write iptable firewall rules, interfaces files, routing rules but I do not know what to do at the network stack level.

All I know right now is when I send pings to the server from another server I get 1-5% packet loss. :-(

Bonno Bloksma


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 89D1798A7351D040B4E74E0A043C69D70107A3F0@HGLEXCH-01.tio.nl">http://lists.debian.org/89D1798A7351D040B4E74E0A043C69D70107A3F0@HGLEXCH-01.tio.nl
 
Old 03-21-2012, 04:12 PM
Camaleón
 
Default packet loss after Lenny Squeeze upgrade

On Wed, 21 Mar 2012 13:21:34 +0000, Bonno Bloksma wrote:

> I have a Debian server (hardware HP DL360) which I use for internal
> routing and which uses NAT to isolate a part of my network. This machine
> has been routing for the past few years without any problems but lately
> there is a lot of packet loss for the traffic going to/through that
> server.
>
> What has changed in Debian that could cause this and what can I tweak to
> get things going again? What do I need to look at?

(...)

The network driver?

As you have updated the system from Lenny to Squeeze, the kernel could
have introduced some sort of incompatibility or bug within the network
module you're using.

I would search for any bug report related to "packet loss" over the
Broadcom Nextreme IIx drivers.

Greetings,

--
Camaleón


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: jkd269$lch$11@dough.gmane.org">http://lists.debian.org/jkd269$lch$11@dough.gmane.org
 
Old 03-22-2012, 01:51 AM
Stan Hoeppner
 
Default packet loss after Lenny Squeeze upgrade

On 3/21/2012 11:00 AM, Bonno Bloksma wrote:

> All I know right now is when I send pings to the server from another server I get 1-5% packet loss. :-(

It's a requirement on a *technical* mailing list to provide technical
data related to a problem, not merely subjective statements, which is
all you have provided thus far. At *minimum* you should provide some
*evidence* of the problem. Such evidence may consist of, but not only,
ping and traceroute output in both directions, to/from the "problem
server", from multiple hosts on different switch/router segments within
your network and outside of it. We need to see your output, not just
comments, such as:

$ ping 217.114.99.194
PING 217.114.99.194 (217.114.99.194) 56(84) bytes of data.
64 bytes from 217.114.99.194: icmp_req=1 ttl=47 time=163 ms
64 bytes from 217.114.99.194: icmp_req=2 ttl=47 time=172 ms
64 bytes from 217.114.99.194: icmp_req=3 ttl=47 time=185 ms
64 bytes from 217.114.99.194: icmp_req=4 ttl=47 time=177 ms
64 bytes from 217.114.99.194: icmp_req=5 ttl=47 time=156 ms
64 bytes from 217.114.99.194: icmp_req=6 ttl=47 time=165 ms
64 bytes from 217.114.99.194: icmp_req=7 ttl=47 time=146 ms
^C
--- 217.114.99.194 ping statistics ---
7 packets transmitted, 7 received, 0% packet loss, time 6023ms
rtt min/avg/max/mdev = 146.948/166.733/185.263/12.009 ms


$ traceroute -q 1 217.114.99.194
traceroute to 217.114.99.194 (217.114.99.194), 30 hops max, 60 byte packets
1 smc.hardwarefreak.com (192.168.100.1) 0.883 ms
2 mo-65-41-216-193.sta.embarqhsd.net (65.41.216.193) 65.722 ms
3 mo-69-68-209-229.sta.embarqhsd.net (69.68.209.229) 66.578 ms
4 208-110-240-186.centurylink.net (208.110.240.186) 68.559 ms
5 bb-kscbmonr-jx9-01-ae0.core.centurytel.net (206.51.69.5) 75.374 ms
6 bb-dllstx37-[...].core.centurytel.net (206.51.69.25) 84.665 ms
7 dal-edge-18.inet.qwest.net (72.165.208.157) 85.815 ms
8 dap-brdr-03.inet.qwest.net (67.14.2.85) 85.794 ms
9 xe-8-1-0.edge2.dallas3.level3.net (4.68.63.49) 90.268 ms
10 vlan90.csw4.Dallas1.Level3.net (4.69.145.254) 92.403 ms
11 ae-83-83.ebr3.Dallas1.Level3.net (4.69.151.158) 94.672 ms
12 ae-3-3.ebr2.NewYork1.Level3.net (4.69.137.122) 124.038 ms
13 ae-63-63.ebr1.Atlanta2.Level3.net (4.69.148.242) 128.148 ms
14 ae-6-6.ebr1.Washington12.Level3.net (4.69.148.106) 126.386 ms
15 ae-1-100.ebr2.Washington12.Level3.net (4.69.143.214) 128.035 ms
16 4.69.148.49 (4.69.148.49) 132.929 ms
17 ae-58-223.csw2.Amsterdam1.Level3.net (4.69.153.210) 220.366 ms
18 ae-2-52.edge4.Amsterdam1.Level3.net (4.69.139.170) 185.791 ms
19 gi1-7.bg1.am1.eu.equinix.net (212.72.40.62) 188.672 ms
20 ae-2-52.edge4.Amsterdam1.Level3.net (4.69.139.170) 188.527 ms
21 ve85.l3c-en1-1.as16243.net (87.249.109.1) 200.309 ms
22 ve1.far4.as16243.net (77.222.66.126) 199.577 ms
23 ve85.l3c-en1-1.as16243.net (87.249.109.1) 192.612 ms
24 ve1.far4.as16243.net (77.222.66.126) 192.505 ms
25 *
26 *
27 *
28 *
29 *
30 *

--
Stan


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 4F6A93AF.3060505@hardwarefreak.com">http://lists.debian.org/4F6A93AF.3060505@hardwarefreak.com
 
Old 03-22-2012, 06:01 AM
frank
 
Default packet loss after Lenny Squeeze upgrade

On Wed, 2012-03-21 at 16:00 +0000, Bonno Bloksma wrote:

> What do you mean with WHERE the loss happens? I know what do to at the "normal" level where I write iptable firewall rules, interfaces files, routing rules but I do not know what to do at the network stack level.

How have you verified that this host is the reason? What reports
traceroute or better mtr?

Do you have still packet loss if you completely disable the firewall?
How does it looks like if you ping from box itself? I assume it's a
switched network, anything in the switch logfile?

What network driver is loaded and what kernel is used? Have you tried to
boot an older kernel?

> All I know right now is when I send pings to the server from another server I get 1-5% packet loss. :-(

How many hops are between the host you have pinged from?

Frank


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 1332399704.25368.10.camel@nero.internal.friendscou t24.de">http://lists.debian.org/1332399704.25368.10.camel@nero.internal.friendscou t24.de
 
Old 03-22-2012, 01:08 PM
Bonno Bloksma
 
Default packet loss after Lenny Squeeze upgrade

Hi,

We have pinned it down even further.
Problem server has multiple nic. Test server is connected to eth0 of problem server.

Using the test server:
Ping the switch to which the problem server is connected and I have no packet loss.
Ping the ip for eth0 and I have no packet loss.
Ping the ip for eth1 and I DO have packet loss.

Seems it is going wrong somewhere after the packet has arrived on the server and gets lost in routing or something like it.

The eth0 NIC uses the Trigon3 driver. Others use the Trigon3 or the Intel Pro 1000 driver.

I cannot just remove the firewall as it is needed for the correct function of the server, it's function is to be a router/firewall.

Switch logs show nothing wrong. Ethernet port speeds/duplex settings have been checked and are ok.

Pinging other machines on the network from the test server shows no packet loss so it seems it is not a network problem but realy a problem on that server itself.

Yours sincerely,
Bonno Bloksma
senior systemadministrator

tio
university of applied sciences

-----Oorspronkelijk bericht-----
Van: frank [mailto:frank@dead-link.org]
Verzonden: donderdag 22 maart 2012 8:02
Aan: debian-user@lists.debian.org
Onderwerp: RE: packet loss after Lenny Squeeze upgrade

On Wed, 2012-03-21 at 16:00 +0000, Bonno Bloksma wrote:

> What do you mean with WHERE the loss happens? I know what do to at the "normal" level where I write iptable firewall rules, interfaces files, routing rules but I do not know what to do at the network stack level.

How have you verified that this host is the reason? What reports traceroute or better mtr?

Do you have still packet loss if you completely disable the firewall?
How does it looks like if you ping from box itself? I assume it's a switched network, anything in the switch logfile?

What network driver is loaded and what kernel is used? Have you tried to boot an older kernel?

> All I know right now is when I send pings to the server from another
> server I get 1-5% packet loss. :-(

How many hops are between the host you have pinged from?

Frank


--
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/1332399704.25368.10.camel@nero.internal.friendscou t24.de


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 89D1798A7351D040B4E74E0A043C69D70107BBBB@HGLEXCH-01.tio.nl">http://lists.debian.org/89D1798A7351D040B4E74E0A043C69D70107BBBB@HGLEXCH-01.tio.nl
 
Old 03-22-2012, 03:17 PM
frank
 
Default packet loss after Lenny Squeeze upgrade

On Thu, 2012-03-22 at 14:08 +0000, Bonno Bloksma wrote:
> Hi,
>
> We have pinned it down even further.
> Problem server has multiple nic. Test server is connected to eth0 of problem server.
>
> Using the test server:
> Ping the switch to which the problem server is connected and I have no packet loss.
> Ping the ip for eth0 and I have no packet loss.
> Ping the ip for eth1 and I DO have packet loss.
>
> Seems it is going wrong somewhere after the packet has arrived on the server and gets lost in routing or something like it.
>
> The eth0 NIC uses the Trigon3 driver. Others use the Trigon3 or the Intel Pro 1000 driver.
>
> I cannot just remove the firewall as it is needed for the correct function of the server, it's function is to be a router/firewall.
>
> Switch logs show nothing wrong. Ethernet port speeds/duplex settings have been checked and are ok.
>
> Pinging other machines on the network from the test server shows no packet loss so it seems it is not a network problem but realy a problem on that server itself.

More information would be helpful. If you feeling uncomfortable posting
the information here, feel free to email them directly to me.

netat -rn

ifconfig

iptables -L


Frank


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 1332433053.9423.3.camel@nero.internal.friendscout2 4.de">http://lists.debian.org/1332433053.9423.3.camel@nero.internal.friendscout2 4.de
 

Thread Tools




All times are GMT. The time now is 01:19 AM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright ©2007 - 2008, www.linux-archive.org