Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations SkipVought on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

IBM Server (8476) looses time or gains time....

Status
Not open for further replies.

dcruz2012

MIS
Feb 19, 2003
8
0
0
US
Please help!!
I have 4 IBM Netfinity servers (8476) which for some reason is either gaining time or loosing time. I have unloaded the SYNCTIME.NLM and re-loaded it to see if this helps(only on 1 server), however, in a span of two days it gained 4 minutes. I also have two of the servers gaining 1 hour; I am suspecting a motherboard issue; however, I've been to IBM web site and even called them, but they did not know anything. I don't know if it makes a difference, but I am running Novell 3.12 on all the servers.
 
First and foremost, Netware v3.x was never tested, and consequently never supported on the Netfinity 3000 by IBM, so it makes sense they could not help. If you are runnin SyncTime.NLM, it is not likely a hardware problem on the server running the .NLM, but the source you are synching with. Start there. Also, you may want to consider updating your OS to a more recent version - Netware 6.x is out.
 
Catorze,
IBM, I think their customer service is a joke; talk about bad support!

Thanks for the input on the question. Let me give some more info on the situation; we are currently updating the servers time from a CISCO router which sync time with approximately 365 other Netfinity servers, only some of these servers are having the time increment problem. I have looked into it geographically and it does not have a pattern - the only pattern is that the servers are all getting this problem one by one. So far out of 365 servers we have about 7 with the problem and this stared (when we were aware of this) about 1 month ago. Changing the OS would be a wonderful idea, however, trying to convince the State to do it, would be another challenge. I appreciate your help and everyone's help in this. It's something that has us dumbfounded...please help me!

Thanks,
Dan.
 
Well several things come to mind.

First, clocks built into computer system boards have always been just a convinence device for the user, accuracy was never a design point.

If accuracy is a requirement your best bet, and most economical, is the free software provided by the National Institute of Standards and Time (NIST) " ". With their software and access the the WWW; you can sync your computers to the Atomic Clock, and you can set a schedule to do it several times a day if you wish.

Back to the system clock. I have worked with small computers from 1977 and have never seen one with a clock that had any acceptable level of accuracy. They first started showing up on a large scale with the old IBM PCAT later known as the ISA bus. The components used just will not support the circuritry for a really accurate clock.

In a nut shell the issue starts with POST. During POST the Real Time Clock (RTC) is read for the current date and time. That info is written to memory and updated as interrupts are generated. All references to time come from memory, not the RTC. So a system that is not seeing a lot of interupts will appear to have a slow clock. You also have the issue of Time zone settings in the OS and delays over the network, etc.

My solution was using the NIST software. It made life a lot easier.

Good luck, Larry
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top