We got 2 CentOS servers on same datacenter. they both have websites using same mysql server. Most of the websites uses same source codes located on a.b.c.162. There is a shortcut to a.b.c.162 on a.b.c.164 placed on same location as a.b.c.162. So websites are reaching source codes through shoutcut between servers if website is located on a.b.c.164.

System was working perfectly until this morning. But know servers can't ping each other but they can ping all the servers on same datacenter and i can ping them from my personal computer.

Any ideas why it happened and how to solve this.


  • Thanks for your answers our network ops couldn't fing the problem they're changing the box :) – skuul Sep 13 '11 at 16:42

Hmm, I would check the arp table on both boxes.

/sbin/arp -an

Make sure the expected IP address matches up with the machines MAC address for the given interface.

In any case, I'd definitely contact network operations about the problem. If there have been no changes on the boxes, then the most likely explanation is that there's an issue at the switch.

| improve this answer | |
  • ? (a.b.c.162) at <incomplete> on eth0 that is the output i got from a.b.c.164 any ideas how i can solve that – skuul Sep 13 '11 at 15:48
  • Not sure, since I believe there are underlying networking issues somewhere else. You can try deleting arp entries, with the "-d" option (check the arp man page for more info), but the arp table should be automatically populated. Again, check with the network ops guys to see if there have been any changes. – cjc Sep 13 '11 at 15:58

General trouble shooting:

  1. verify that all the machine have the expected IP address and netmask with ifconfig -a
  2. while you're looking at the output of ifconfig, check for errors on the interface
  3. verify that the default route is correct, and that there are no other wrong routes in your routing tables with netstat -r
  4. verify that the arp entries for the machines you can't ping are not wrong with arp -n
  5. if you're using hostnames rather than IPs to try to communicate, use nslookup to check that the DNS entries are returning correctly

If everything looks right on all the servers, then it's time to contact the person who manages the switches in the data centre to make sure that the managed switches are properly configured. Sometimes it helps to clear the arp cache on the relevant routers/switches, and sometimes it's a simple case of a switch being rebooted and forgetting it's config.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.