Amanda-Users

Re: How to determine cause of Amanda slowdown?

2002-08-08 13:04:27
Subject: Re: How to determine cause of Amanda slowdown?
From: "Paul G. Allen" <pgallen AT randomlogic DOT com>
To: amanda-users AT amanda DOT org
Date: Thu, 08 Aug 2002 09:54:25 -0700
KEVIN ZEMBOWER wrote:
> 
> Thank you to all who responded, especially Christopher, Jousha and Mike.
> Most of you suggested checking the NIC and switchport for mismatches.
> 
> This is a portion of "ifconfig eth0":
> admin:/proc/net # ifconfig eth0
> eth0      Link encap:Ethernet  HWaddr 00:90:27:B6:FB:E7
>           inet addr:172.16.2.7  Bcast:172.16.255.255  Mask:255.255.0.0
>           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>           RX packets:259376827 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:188989102 errors:0 dropped:0 overruns:0
> carrier:65070
>           collisions:54220248 txqueuelen:100
>           RX bytes:2574728265 (2455.4 Mb)  TX bytes:2729800656 (2603.3
> Mb)
> 
> This is the system on the 100Base-T line connected directly to a Cisco
> switch, which should be set at full duplex. Does the fact that they're
> any collisions at all indicate that this is not working in full duplex
> mode? The collisions are 21% of the number of RX packets, and 28% of the
> TX packets. Are these numbers excessive?
> 

IIRC, we had many problems with CISCO switches and NICs at Akamai that
were supposed to be at full duplex and were not. Some we had to set a
half duplex on both the switch and the NIC, some we could get working at
full by messing around with either/both. I couldn't tell you what brands
NICs and switch models were the culprits (someone else had the actual
job of fixing these problems), but it might be something to consider. In
ever case though, we could not make a configuration work if switch and
NIC were both set to auto - every NIC and switch port in the place that
I knew of was specifically set to 100mb and either full or half duplex.

And just for comparison, I am running on an auto sensing full-duplex
10/100 switch (a Netgear RO318) and NIC (MoBo based 3Com, Tulip chipset,
on this system), and here's my ifconfig. Note the lack of collisions:

[root@keroon root]# ifconfig
eth0      Link encap:Ethernet  HWaddr 00:E0:81:03:80:C7  
          inet addr:192.168.1.4  Bcast:192.168.1.255  Mask:255.255.255.0
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:53078505 errors:0 dropped:0 overruns:0 frame:0
          TX packets:76216654 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:100 
          RX bytes:3278523039 (3126.6 Mb)  TX bytes:862193510 (822.2 Mb)
          Interrupt:18 Base address:0x1800 

lo        Link encap:Local Loopback  
          inet addr:127.0.0.1  Mask:255.0.0.0
          UP LOOPBACK RUNNING  MTU:16436  Metric:1
          RX packets:463 errors:0 dropped:0 overruns:0 frame:0
          TX packets:463 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0 
          RX bytes:60113 (58.7 Kb)  TX bytes:60113 (58.7 Kb)

[root@keroon root]# 


I also have a Linksys 10/100 switch that gives the same results
(currently not in use).

PGA
-- 
Paul G. Allen
Owner, Sr. Engineer, Security Specialist
Random Logic/Dream Park
www.randomlogic.com