Hi All,

I have now spoken to so many people and yet nobody have been able to provide me with an answer on this so I am just taking a chance here.

currently we have 12 Linux servers which were configured to sync the time on the serverd using NTP.

All 12 servers are running RHEL (4 X rhel4.5 and 8 X rhel5.4).

We have been experiencing a lot of issues on our servers in terms of the times going out of sync and as they are running databases in a rac/cluster environment the times are not allowed to be out of sync as this breaks db functionality.

after trying many changes, we still have the following issue:

- All 12 servers are manually synced to the time server and the times are correct on all the servers.

- after about a week though, we will start noticing that all the times on the servers differ from one another by about 1minute

- after about a month, the time will be out by 5-10 minutes and then start to produce problems on the db side.


We have tested the servers and they can all see (ping) the time server using both IP and hostname.


Any suggestions on what might be going on will be greatly appreciated.

Regards
Machiel

PS.... Looking in the /var/log/messages file no errors are found that the ntp sync has failed and for that matter none to suggest it was synced either.

ntpstats simply says that the time is in sync by up to 11ms, however this is not the case and the time server's time is different from all 12 servers.

you should check "ntpq -pn" on the relevant servers to see the current sync state. It's common to get issues with NTP on VM's but if these are physical then it really should just be poor ntp syncing. How many ntp servers are you using? Should be at least 3, but preferably many more.