Linux Server FAQ

General class hint eth1: Too much work at interrupt, IntrStatus=0x0001 The meaning of this hint is. There are too many interrupt requests for a NIC. If it happens only once, it can be ignored. Tips If there are frequent occurrences or concentrated episodes, there are more possibilities that may need to be addressed. More possibilities, such as network card performance; server performance; network attacks.. and so on. General class hints IPVS: incoming ICMP: failed checksum from 61.172.0.X! The server received an ICMP packet with a checksum error. This type of data packet may be illegally generated garbage data. But from the current point of view, the server receives such data very much. It is generally ignored. A typical proxy server will forward thousands of packets per second while it is working. Receiving several error packets will not affect normal work. This is the most common type of prompt. The general class prompt NET: N messages suppressed. The server ignores N packets. Similar to the previous one. The packets received by the server are considered to be useless garbage data. This type of data is mostly generated by the attack class program. of. This hint can be ignored if N is small. However, if there are such prompts of more than 3 digits of data for a long time or longer, it is very likely that the server is attacked by the bandwidth of the garbage data class. General class prompt UDP: bad checksum. From 221.200.XX:50279 to 218.62.XX:1155 ulen 24 UDP: short packet: 218.2.XX:3072 3640/217 to 222.168.XX:57596 218.26.131.X sent an invalid ICMP type 3, code 13 error to a broadcast: 0.1.0.4 on eth0 The server received an error packet. UDP checksum error; too short UDP packet; a wrong ICMP type data. It is also illegally produced under normal circumstances. But the general problem is not directly ignored. General class hints kernel: conntrack_ftp: partial 227 2205426703+13 FTP_NAT: partial packet 2635716056/20 in 2635716048/2635716075 The server has an error while maintaining a connection to the FTP protocol. Such prompts can generally be ignored. Serious network communication problem! NETDEV WATCHDOG: eth1: transmit timed out eth1: link down eth1: link up, 10Mbps, half-duplex, lpa 0x0000 eth2: link up, 100Mbps, full-duplex, lpa 0x41E1 setting full-duplex based on MII #24 link partner capability of 45e1 These prompts appear when there is a serious problem in network communication. The fault is basically related to the network disconnection. The meanings of these prompts are that a certain network card transmits data timeout; the network card is connected down; The network card is connected up, the connection rate is 10/100Mbps, full/half duplex. The prompts of the last three lines written here are similar. When this kind of prompt appears, you must pay attention to the network connection status to handle!!! Network communication serious problem! NIC Link Is Up 100 Mbps Full Duplex case and kernel: eth1: link up, ... the same. It refers to the connection rate of a certain network card. It is generally considered that there is no description of which network card is down, but the continuous adaptation of the network card rate is also a problem with communication. If the normal disconnection of the network cable can ignore such information. Serious problem with network communication! eth0: Transmit timed out, status 0000, PHY status 786d, resetting... eth0: Reset not complete yet. Trying harder. The first prompt fails to send data to the network card. Reset the network card. The second prompt network card Reset is unsuccessful.... These tips are serious communication problems. Alarm program prompt 0001 ##WMPCheckV001## 2005-04-13_10:10:01 Found .(ARP Spoofing sniffer)! IP:183 MAC:5 0002 ##WMPCheckV001## 2005-04-07_01:53:32 Found . (MAC_incomplete)! IP:173 mac_incomplete:186 0003 ##WMPCheckV001## 2005-04-17_16:25:11 Found .(HIGH_synsent)! totl:4271 SynSent:3490 0004 ##WMPCheckV001## 20...... This is a reminder caused by the alarm program. Detailed information needs to be received in real time by the client of the alarm program. For details, please see the "Alarm module and log". General class prompt eth1: Promiscuous mode enabled. device eth1 entered promiscuous mode device eth1 left promiscuous mode These lines indicate that a certain network card enters (leaves) the promiscuous mode. Generally, the promiscuous mode is when the communication needs to be captured. Used only when using maintenance or failure analysis (such as the countflow command in consoletools). Normally generated such prompts can be ignored. This prompt should be noticed if there is no maintenance in the foreground and at the far end, but this is unlikely. Basically irrelevant keyboard: unknown scancode e0 5e The keyboard receives an undefined key value. If it occurs frequently, there may be a problem with the keyboard. Linux sometimes has such a hint for a special key or a combination key. It's up to you to see if the server's keyboard is pinched. Other things are generally ignored. Basically irrelevant uses obsolete (PF_INET, SOCK_PACKET) The system kernel calls some of the function modules, which will appear the first time you call in. The general situation is related to the use of debugging tools. Can be ignored directly. Network communication failure Neighbour table overflow. This prompt appears. It is generally because some computers in the LAN are infected by viruses. In severe cases, communication will be affected. Computers with abnormal internal network communication must be handled. Network communication failure eth1: Transmit error, Tx status register 82. Probably a duplex mismatch. See Documentation/networking/vortex.txt Flags; bus-master 1, dirty 9994190(14) current 9994190(14) Transmit list 00000000 vs. f7171580. 0: @f7171200 length 800001e6 status 000101e6 1: @f7171240 length 8000008c status 0001008c .... This prompt is unique to the 3com network card. I feel that if the amount is not large, it will not affect the serious. The current solution to look at the first one is Replace the network card on the server. I really feel that there are some problems with 3com's network card...

Server system severe fault CPU0: Temperature above threshold CPU0: Running in modulated clock mode Server CPU operating temperature is too high. Hardware failure must be ruled out.