I don't believe the sysplex timers are "ntp-capable". The external time reference connects via RS-232, but here's an interesting link for how to make your own etr...
http://www.share.org/proceedings/sh96/data/S2968.PDF Personally I'm not sure I'd trust my crown jewels to such a configuration but it does sound quite intriguing. Scott Chapman |---------+----------------------------> | | "McKown, John" | | | <[EMAIL PROTECTED]| | | tr.com> | | | Sent by: Linux on| | | 390 Port | | | <[EMAIL PROTECTED]| | | IST.EDU> | | | | | | | | | 06/02/03 08:54 AM| | | Please respond to| | | Linux on 390 Port| | | | |---------+----------------------------> >--------------------------------------------------------------------------------------------------------------------| | | | To: [EMAIL PROTECTED] | | cc: | | Subject: Re: Time Sync | >--------------------------------------------------------------------------------------------------------------------| No. The ONLY thing that will sync the mainframe's clock is an ETR (External Time Reference) aka "sysplex timer". Now, Linux and VM may not care about the time being "fiddled with". But OS/390 and z/OS are practically guaranteed to die a horrible death if the TOD clock "goes backwards". The architecture defination specifically states that the TOD clock is strictly increasing in value. The sysplex timer adjusts the timer backwards by slowing the TOD clock down until it matches the value on the sysplex timer. I think that I've read that the sysplex timer can use another "time source", but I don't know if it is ntp capable or not. Note that the OS/2 service element is the source of the mainframe's TOD clock *ONLY* when a POR is done. If you like to live very dangerously and in an unsupported way, look at the SCK instruction. However, if you have an MVS system running under VM and use VM to actually change the TOD clock, you can expect MVS to die a horrible death. But likely only have having DB2, IMS and CICS problems. All these systems assume that the TOD clock never goes "backwards". -- John McKown Senior Systems Programmer UICI Insurance Center Applications & Solutions Team +1.817.255.3225 This message (including any attachments) contains confidential information intended for a specific individual and purpose, and its' content is protected by law. If you are not the intended recipient, you should delete this message and are hereby notified that any disclosure, copying, or distribution of this transmission, or taking any action based on it, is strictly prohibited. > -----Original Message----- > From: Steven A. Adams [mailto:[EMAIL PROTECTED] > Sent: Friday, May 30, 2003 5:36 PM > To: [EMAIL PROTECTED] > Subject: Re: Time Sync > > > On Fri, 2003-05-30 at 14:10, Alan Cox wrote: > > On Gwe, 2003-05-30 at 22:40, Steven A. Adams wrote: > > > This might be a little off topic so feel free to let me > know if it is. > > > > > > Our z/800 is about 90 seconds off of time with the rest > of the network > > > and this is starting to cause some havoc with an > application that we are > > > currently developing. I have been told that getting the > mainframe to > > > sync up is probably an expensive proposition. I am being > told that there > > > is not a way to use ntpd under one of my Linux guests to > set the clock > > > and propagate these settings through VM to the CTC. So, > if you don't > > > mind me asking, what solutions are commonly used for time sync? > > > > In the Linux world and to an extent nowdays in the Windows world NTP > > seems to be the popular system. Its an internet standard > that will let > > you lock your machiens to each other and to global time sources that > > ultimately anchor back to atomic clocks. > > > > xntpd is the Linux daemon for this, but I don't know if VM > has anything > > of its own to talk ntp > > Thanks Alan, > I am very familiar with ntp. My question was intended to see > what others > do to sync the hardware clock on the mainframe itself. Is there an ntp > equivalent that will set and maintain the hardware clock on the z/800? > -- >
