Re: [ilugd] Re: Time Lag in Linux...
My machine is a stand alone one and is not connected to LAN or internet.. SO i cannot use a time server to sync the time Also i changed the CMOS battery only recently Well, worst-case, I've seen faulty motherboards which have a malfunctioning clock. On one or two occasions, the only way we could fix it was by replacing the motherboard. Shuvam ___ ilugd mailing list [EMAIL PROTECTED] http://frodo.hserus.net/mailman/listinfo/ilugd
Re: [ilugd] Re: Time Lag in Linux...
LinuxLingam wrote: it is also not necessary for the pc to be 'old' for the CMOS battery to fail. other possible reasons; 1) CMOS battery drained if its a recycled/recharged type peddled by some shady guy. 2) drained due to a shortage or leakage. I changed the CMOS battery recently!!! 3) if you have those horrendous PCs that never really poweroff, as in HARD OFF. they tend to look like they've all shut down, but one press on the keyboard and they can powerup again, called a soft on. well, if you have the sense of physically pulling out the power cable from the socket so the PC won't fry while its offically switched off (happens far more frequently than you can imagine) the CMOS battery gets drained more rapidly. nevertheless, i never expect any CMOS-based timing system to be 'mission critical' on the usual i386-type architecture. as ghane rightly points out, you need serious hardware for that kind of thing, or an always on, high-speed OK... some background about my application We develop s/w and hardware for a small, indegeneously developed telephone exchange . The main call processing s/w runs on a i386 PC which is connected to the exchange through some properitary h/w to the main exchange. All the events are controlled and triggered through our s/w. We selected linux as OS long time back when linux was in its infancy.. This PC is not connected to external world by any means.. As this call processing s/w does all the main functinalities.. it is absolutely necessary to sync the time to real time. And also the application is expected to run for days along continusoly So i cannot use any other time server to sync the time Any other input...!!! ./zahoor ___ ilugd mailing list [EMAIL PROTECTED] http://frodo.hserus.net/mailman/listinfo/ilugd
Re: [ilugd] Re: Time Lag in Linux...
On Tuesday 19 August 2003 02:50 pm, you wrote: I changed the CMOS battery recently!!! OK... some background about my application [snip] by any means.. As this call processing s/w does all the main functinalities.. it is absolutely necessary to sync the time to real time. And also the application is expected to run for days along continusoly So i cannot use any other time server to sync the time Any other input...!!! ./zahoor must confess i am way out of my depth here. apart from the usual suggestions given by everybody else (1) use the latest hardware, (2) use the latest version of the kernel, optimized for your requirements. strange, i have never really seen a test or a diagnostic tool that tests to see *if* the time-clock circuits on a motherboard are accurate, and how accurate. so use the idential software soup but on a different motherboard, all else equal, and see if it is a peculiar motherboard-related problem only. anybody else has a solution to this problem? ___ ilugd mailing list [EMAIL PROTECTED] http://frodo.hserus.net/mailman/listinfo/ilugd
Re: [ilugd] Re: Time Lag in Linux...
Why don't you just get yourself a RTC card and use that instead of relying on the builtin clock!! On Tue, 19 Aug 2003 14:50:52 +0530, J.Mohamed Zahoor [EMAIL PROTECTED] wrote: LinuxLingam wrote: it is also not necessary for the pc to be 'old' for the CMOS battery to fail. other possible reasons; 1) CMOS battery drained if its a recycled/recharged type peddled by some shady guy. 2) drained due to a shortage or leakage. I changed the CMOS battery recently!!! 3) if you have those horrendous PCs that never really poweroff, as in HARD OFF. they tend to look like they've all shut down, but one press on the keyboard and they can powerup again, called a soft on. well, if you have the sense of physically pulling out the power cable from the socket so the PC won't fry while its offically switched off (happens far more frequently than you can imagine) the CMOS battery gets drained more rapidly. nevertheless, i never expect any CMOS-based timing system to be 'mission critical' on the usual i386-type architecture. as ghane rightly points out, you need serious hardware for that kind of thing, or an always on, high-speed OK... some background about my application We develop s/w and hardware for a small, indegeneously developed telephone exchange . The main call processing s/w runs on a i386 PC which is connected to the exchange through some properitary h/w to the main exchange. All the events are controlled and triggered through our s/w. We selected linux as OS long time back when linux was in its infancy.. This PC is not connected to external world by any means.. As this call processing s/w does all the main functinalities.. it is absolutely necessary to sync the time to real time. And also the application is expected to run for days along continusoly So i cannot use any other time server to sync the time Any other input...!!! ./zahoor ___ ilugd mailing list [EMAIL PROTECTED] http://frodo.hserus.net/mailman/listinfo/ilugd -- Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/ ___ ilugd mailing list [EMAIL PROTECTED] http://frodo.hserus.net/mailman/listinfo/ilugd
Re: [ilugd] Re: Time Lag in Linux...
OK... some background about my application We develop s/w and hardware for a small, indegeneously developed telephone exchange . The main call processing s/w runs on a i386 PC which is connected to the exchange through some properitary h/w to the main exchange. All the events are controlled and triggered through our s/w. We selected linux as OS long time back when linux was in its infancy.. This PC is not connected to external world by any means.. As this call processing s/w does all the main functinalities.. it is absolutely necessary to sync the time to real time. And also the application is expected to run for days along continusoly Now that it is clear that real-time clocking is a mission-critical business necessity, without which your actual business data would go haywire, it seems to me that someone somewhere goofed _badly_ by choosing an off-the-shelf Intel base for this project. Sorry if it sounds blunt, but that's the way I see it. At the very _least_ this project should have been on SPARC or some such better-quality hardware. All of a sudden, Ghane's SPARC suggestion does not seem like a joke at all. Now that you can't undo what has been done, I suggest that you evaluate external time clock hardware (which have Linux and NTP support) and plug some such hardware to the serial port of the PC. Such hardware is described in NTP related literature. Many options exist, including radio receivers which receive time signals from super-accurate clocks, and plain super-accurate, industrial strength external RTC hardware. And if none of that works, get a second PC, connect it to the Internet using a modem (I'd suggest a Reliance mobile phone), and run an NTP client on it. Make it connect to the Net for, say, 10 minutes every two hours, and sync its clock with NTP servers elsewhere. And get your existing telephone exchange to talk to this second PC and get its clock in sync. This is inelegant, but may be easiest to strap together with string and bandage if nothing else works. Shuvam ___ ilugd mailing list [EMAIL PROTECTED] http://frodo.hserus.net/mailman/listinfo/ilugd
Re: [ilugd] Re: Time Lag in Linux...
On Tuesday 19 August 2003 12:13 am, you wrote: Also, just asking, ... is your machine an old one? In which the CMOS battery needs replacement? :) - Sandip it is also not necessary for the pc to be 'old' for the CMOS battery to fail. other possible reasons; 1) CMOS battery drained if its a recycled/recharged type peddled by some shady guy. 2) drained due to a shortage or leakage. 3) if you have those horrendous PCs that never really poweroff, as in HARD OFF. they tend to look like they've all shut down, but one press on the keyboard and they can powerup again, called a soft on. well, if you have the sense of physically pulling out the power cable from the socket so the PC won't fry while its offically switched off (happens far more frequently than you can imagine) the CMOS battery gets drained more rapidly. nevertheless, i never expect any CMOS-based timing system to be 'mission critical' on the usual i386-type architecture. as ghane rightly points out, you need serious hardware for that kind of thing, or an always on, high-speed connection on a high-speed machine doing ntp. also, i have never really quite understood what is it with operating systems, for instance, why is QNX a true 'realtime' operating system, while gnu/linux isn't? or many of the high-end unix variants on big iron machines? so your OS needs to be a true realtime OS as well, depending on your tolerance threshold of what is realtime enough. for this, try QNX, or the truly free TRON operating system. :-) LL ___ ilugd mailing list [EMAIL PROTECTED] http://frodo.hserus.net/mailman/listinfo/ilugd
Re: [ilugd] Re: Time Lag in Linux...
On Tuesday, August 19, 2003 8:50 AM [GMT+0800], Shuvam Misra [EMAIL PROTECTED] wrote: You just need to find appropriate NTP servers which work for you, that's all. I guess junta on this list can help there. pool.ntp.org -- Sanjeev ___ ilugd mailing list [EMAIL PROTECTED] http://frodo.hserus.net/mailman/listinfo/ilugd
Re: [ilugd] Re: Time Lag in Linux...
My machine is a stand alone one and is not connected to LAN or internet.. SO i cannot use a time server to sync the time Also i changed the CMOS battery only recently ./zahoor Sandip Bhattacharya wrote: As ghane said, use ntpd or xntpd - whatever is available on your distribution. Also, just asking, ... is your machine an old one? In which the CMOS battery needs replacement? :) - Sandip ___ ilugd mailing list [EMAIL PROTECTED] http://frodo.hserus.net/mailman/listinfo/ilugd