[EMAIL PROTECTED] wrote: > Hi All, > > I am currently at a site where we have workstations running an > application program that requires accurate time. The application > writes a date time stamp to a database, the dts is generated locally on > the machine. Another applcation periodically copies data from the > original database, to another database querying by the dts since it > last ran. This is why the time needs to be accurate on all clients, > please note though, that the dts resolution is only down to the second. >
You can modified your stored procedures so that the database provides the timestamp rather than the workstations. > All the client workstations currently use xntpd for time keeping, and > we have a time server etc. all already configured, thing is we seem to > still be getting issues where clients aren't having their time adjusted > correctly, or maybe quickly enough ? You need to figure out why not. How do you know that that is what is happening? Are there any firewalls? If you are running Linux, newer versions of the O/S by the vendors turn on firewalls by default. > > I think NTP is just overkill for this site, On the contrary, based on your description above, if you want the workstations to provide the timestamp you *NEED* NTP. Nothing else would provide you with the accuracy you need. I haven't done a full > in-depth investigation into the NTP RFC's but it seems it is a very > sophisticated time keeping protocol, and it has an algorithm that > doesn't just slam the time back in line upon finding clock drift, also > I have read that there maybe a maximum drift for which the protocol > will adjust. You shouldn't worry about the algorithms, they've been developed over years of experimentation and practice. The drift is something you should only need to deal with upon startup. > > We have hundreds of workstations and sometimes they may have been > switched off for a peroid of time, or the CMOS battery may have died - > does this affect NTP on the whole ? > No, just used the ntpd -g option in the startup and it will reset the time to the correct time. The -g option allows it to make large corrections the first time after which ntpd corrects the clock more finely. > Basically, could someone give me a brief primer on NTP from a > protocol/algorithm/operational point of view (if possible) > If you really want to see the technical documentation on the algorithms and implementation you should go to the ntp.org web site here: http://www.eecis.udel.edu/~mills/ntp.html > I think we could be better with just a very basic SNTP ?, or just > UDP/TIME implementation > No, not based on your requirements. > All we need is for the machine's time to be brought in line > peroidically accurate to the second from a reference machine. > NTP does much more than this, but it has almost no effect on the running system so why worry about installing it. > Unfortunately we can't do this simply with login scripts because a) we > don't have desktop permissions to change time, and b) users could be > left logged in for days at a time, so we still need some sort of active > time management. > NTP requires the privilege to change the time as would any method you would use to change the time. Danny _______________________________________________ questions mailing list [email protected] https://lists.ntp.isc.org/mailman/listinfo/questions
