ifconfig output explain

aamerjavaid · 10-27-2011, 02:50 AM

Hi All,
I am from the db team, and we are having some performance issues with application, we suspect it might be related with TCP or network stack,

Can someone please explain me the following output of ifconfig, and what needs to be done to rectify this.

Code:

[root@db ~]# ifconfig
eth0      Link encap:Ethernet  HWaddr 3C:4A:92:74:9B:A8
          inet addr:165.1.1.111  Bcast:165.1.255.255  Mask:255.255.0.0
          inet6 addr: fe80::cbf8:92ff:fe74:9ba8/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:144306942 errors:6163 dropped:0 overruns:0 frame:6163
          TX packets:199746835 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000
          RX bytes:58554097 (55.8 MiB)  TX bytes:2955776115 (2.7 GiB)
          Interrupt:169 Memory:f4000000-f4012100

lo        Link encap:Local Loopback
          inet addr:127.0.0.1  Mask:255.0.0.0
          inet6 addr: ::1/128 Scope:Host
          UP LOOPBACK RUNNING  MTU:16436  Metric:1
          RX packets:567858396 errors:0 dropped:0 overruns:0 frame:0
          TX packets:567858396 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:950555498 (906.5 MiB)  TX bytes:950555498 (906.5 MiB)

Thanks,
Amir

fukawi1 · 10-27-2011, 03:12 AM

What specifically do you want explained?

What do you think needs rectifying?

aamerjavaid · 10-27-2011, 04:02 AM

Hi fukawi1,

Thanks for update, Actually we are having some Oracle waits like "Sql*net message from client/Sql*net message to client", so it looks like it might be related with TCP or network stack, which might be working poorly or misconfigured, so that Oracle has to wait for these events.

Please take a look at errors/txqueuelen/packets/Interrupt/etc, do you think it is working fine, or whether we can/should adjust MTU so that oracle don't have to wait for such events.

Thanks,
Amir

jschiwal · 10-27-2011, 09:29 PM

Is your database traffic across the Internet? The IP address for eth0 is a public address. The class-B network is owned by the Associated Press.

I think the frames: field indicates invalid received frames. You may need to examine traffic with a tool such as wireshark to examine what traffic is corrupt, and where it is coming from. I doubt changing the MTU would help, unless a router is breaking up frames to smaller units, and parts of them are being lost.

aamerjavaid · 10-28-2011, 05:03 AM

Hi jschiwal,

Database traffic is local but is internet based using erp system (given below), I have to change IPs before posting. Ok. i will check wireshark and try to check router.

Quote:

[root@erp ~]# ifconfig
eth0 Link encap:Ethernet HWaddr 00:1C:66:07:E6:3F
inet addr:165.1.1.112 Bcast:168.1.1.255 Mask:255.255.0.0
inet6 addr: 2002:a801:2c2:b:21a:64ff:fe07:f63e/64 Scope:Global
inet6 addr: fec0::b:21a:64ff:fe07:f63e/64 Scope:Site
inet6 addr: fe80::21a:64ff:fe07:f63e/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:757219446 errors:6163 dropped:0 overruns:0 frame:6163
TX packets:777261279 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:872940292 (832.5 MiB) TX bytes:1348395248 (1.2 GiB)
Interrupt:169 Memory:ce000000-ce012100

eth1 Link encap:Ethernet HWaddr 00:1C:66:07:E6:3F
inet addr:202.125.140.143 Bcast:202.125.140.191 Mask:255.255.255.192
inet6 addr: 2002:a801:2f2:b:21c:64aa:fc07:b640/46 Scope:Global
inet6 addr: fab0:a801:b:21c:64aa:fc07:b640/46 Scope:Site
inet6 addr: fe80::21f:64af:ce07:f640/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:218543099 errors:6163 dropped:0 overruns:0 frame:6163
TX packets:80643708 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:779268078 (743.1 MiB) TX bytes:1221807321 (1.1 GiB)
Interrupt:209 Memory:ca000000-ca012100

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:110087333 errors:0 dropped:0 overruns:0 frame:0
TX packets:110087333 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:1524480538 (1.4 GiB) TX bytes:1524480538 (1.4 GiB)
[root@erp ~]#

Thanks,
Amir

jschiwal · 10-28-2011, 08:12 AM

It may be a problem with a client as well. If the traffic is local, you shouldn't be seeing the RX errors. I'm assuming you have switches and not hubs.

I just mentioned routers, because on an internet connection, you go through a lot of them, and you asked about MTU. You need to diagnose where the problem is, and not make hit and miss changes.