Many well-known books by Thomas Limoncelli on system administration, very valuable for the admin books. On his
site, too, a lot of interesting, among other things, found such an article:
"A list of dumb things to check" . In Russian, unfortunately, she did not google, she translated herself. After doping, I will use it as part of the KMB in my department, and post the initial translation here.
I didn’t master some moments (I didn’t have enough language or technical knowledge), but I think it suits for practical use. Author's punctuation is partially preserved.
The article (and his entire blog) is distributed under CC-BY-NC-SA.
"
You solve a problem for many hours and suddenly realize: “I bet on the matter in some nonsense!” Often it is. Therefore, we represent ...
')
check list of stupid things
Added: 2009-11-25: People refer to this as something funny. Damn, this list is 100% true! All these things happened to me and made me think: “I'd rather write down all this, so as not to forget anything!” Even the last point!
Level 0 - Human
1. Make sure CapsLock is turned off. (Same for ScrollLock and NumLock)
2. Print it again (without copy-paste) and see if the result is the same. (helps from typos) (and from the whitespace unicode character)
3. Copy and paste the name of the variable (or URL, commands, etc.) to make sure it is entered correctly.
4. You run those programs, what do you think? (Maybe you installed them in single user mode, when / opt was not mounted? Can you compare the checksums of the files of this machine and the machine that works correctly?)
5. Check file permissions.
6. Is this really the computer that you think it is?
7. Do you perform validation with the right machine? Maybe it will be more effective with another?
8. Does a validation check really verify what you think? What if we perform a check of a known-good and obviously defective systems? Did you get the results you expected?
9. Is it a file, directory, hard link, symbolic link or mount point?
10. Is the file extension correct? Maybe it should be .php, not .html?
11. Is the screen stopped by pressing Ctrl-S? (Press Ctrl-Q to find out.)
12. Can you get on the website? Do you work offline?
Level 1 - Physical
1. Are both ends of the power / network / video / ... cable connected?
2. Is the cable plugged into the correct connector?
3. Did you try to disconnect the cable and turn it on again to make sure it is ok?
4. The question "Is it connected?" Is very simple to answer "Yes." It is better to ask to turn it off and on again, or to check both ends of the power cable, or to ask whether the power indicator is on, not lit or flashing.
5. If you turn on the lamp in this outlet - will it light up?
6. Is it a laptop that runs on battery? (and, thus, does not notice the problem with electricity)
7. Is the driver you are installing really from your device?
Level 2 - Channel
1. Is there an outside DHCP server on the network?
2. Is the address pool on the DHCP server exhausted, are there any free addresses left?
3. Does the VPN connection terminate at arbitrary moments, and does the server receive an ICMP response “host unreachable”? The client computer firewall blocks ARP, and after the storage of the ARP cache (300 seconds), the router cannot send packets to it. (Some manufacturers keep an ARP address record all the time while they receive packets from this address (Cisco), but most do not) (I added this because I met this in two places)
Level 3 - Network
1. Is the default gateway specified? Is the network mask set correctly?
2.
http://www.psc.edu/~mathis/MTU/index.html : six classes of problems that reduce network performance:
* packet loss, damage, congestion, faulty equipment
* IP routing, high RTT (round trip time)
* reordering of packages
* inappropriate buffer space
* inappropriate package size
* inefficient applications
3. Does this ip address have someone else? (Unplug the network cable and ping the address)
4. (problems with firewall rules) Does the problem remain after changing the IP address? The problem remains after moving to another subnet? The problem remains if you give this IP address to another machine? The problem remains if you boot another OS (for example, Linux or FreeBSD from a CD-ROM)?
5. Will the same problem arise if you specify an IP address instead of a machine name?
Level 4 - Transportation
1. Trace from A to B. Then from B to A. Do they match each other?
Level 5 - Session
1.
SSH, SCP, L2TP, PPTP
Level 6 - Executive
1. Does the program read the last line of the file? Is it properly processed?
2. Is there a (invisible) Ctrl-M at the end of each line of the text file?
3. Does the file end with a line break?
4. If the line ends with an escape character "\", does it really end there? Is there an invisible space or tab after it? An invisible space or tab can lead to a "unexpected |" shell error!
Level 7 - Applied
1. Is the DNS configured correctly? The error in the DNS settings hides other problems and looks like a problem whose solution you will look for anywhere except /etc/resolv.conf
2. Check environment variables.
3. Do you run the copy of the script that you are editing?
4. The program is looking for a configuration file in the same place as you? (Maybe the new version looks for it at /etc/example2/example.conf instead of the old address /etc/example.conf?)
Level 8 - User / Political
1. Do users actually press RETURN when you guess? (Do they even press it?)
2. The user writes "/" or "\"?
3. Does the user know what the symbols "less" (<) and "more" (>) are?
4. “Do you have permission to crack this password file?” ... “Is it in writing?”
5. Today is the first day of the month? Maybe a problem with billing, and something turned off.
6. Accurately measure time, how often the problem occurs, or after what time it occurs. Mark off on timeline. What protocol can they be associated with?
ARP caches are stored for 300 seconds. Routing protocols are usually updated every 5 or 15 minutes. Once I built a schedule of “pauses in the network” and saw that they happen every 10 minutes, but not only: the router's CPU was overloaded with a process that started every 10 minutes And every time during the update of the RIP. Without a schedule, I would notice only "every 10 minutes" -component.
If nothing helped
Did you remember to check this list?
Tom Limoncelli
"
Original