| LinHES Forums http://forums.linhes.org/ |
|
| Severe Time Drift on MSI K8N Neo4 MBoard http://forums.linhes.org/viewtopic.php?f=2&t=14242 |
Page 1 of 2 |
| Author: | seano [ Sat Feb 17, 2007 3:23 pm ] |
| Post subject: | Severe Time Drift on MSI K8N Neo4 MBoard |
Hello, I'm getting some pretty nasty time drift on my MSI K8N Neo4 Platinum Socket 939 NVIDIA nForce4 Ultra ATX AMD Motherboard. Here's what I've done to "resolve" the issue: - lilo.conf Code: append="splash=silent apm=power-off nomce quiet vmalloc=256m noapic noioapic" - crontab Code: */5 * * * * /usr/sbin/ntpdate time-a.nist.gov >>/var/log/mythtv/ntpdate.log 2>&1 - checked my bios for any "spread spectrum" -- there were 3 and all were disabled (the 3 were for CPU, SATA, PCIE) Looking at my ntpdate log it appears that I'm still getting a pretty significant drift: Code: 17 Feb 12:15:00 ntpdate[3987]: adjust time server 129.6.15.28 offset -0.036796 sec 17 Feb 12:20:02 ntpdate[4346]: adjust time server 129.6.15.28 offset -0.036349 sec 17 Feb 12:25:02 ntpdate[4635]: adjust time server 129.6.15.28 offset 0.005122 sec 17 Feb 12:30:02 ntpdate[4917]: adjust time server 129.6.15.28 offset -0.012068 sec 17 Feb 12:35:02 ntpdate[5209]: adjust time server 129.6.15.28 offset 0.002535 sec 17 Feb 12:40:01 ntpdate[5498]: adjust time server 129.6.15.28 offset -0.024420 sec 17 Feb 12:45:02 ntpdate[5795]: adjust time server 129.6.15.28 offset 0.002944 sec 17 Feb 12:50:02 ntpdate[6103]: adjust time server 129.6.15.28 offset -0.019036 sec 17 Feb 12:59:53 ntpdate[3992]: step time server 129.6.15.28 offset -8.589862 sec 17 Feb 13:00:01 ntpdate[4129]: adjust time server 129.6.15.28 offset -0.202543 sec 17 Feb 13:05:00 ntpdate[4414]: step time server 129.6.15.28 offset -1.884823 sec 17 Feb 13:10:02 ntpdate[4702]: adjust time server 129.6.15.28 offset -0.152578 sec 17 Feb 13:15:02 ntpdate[4993]: adjust time server 129.6.15.28 offset -0.191208 sec 17 Feb 13:20:02 ntpdate[5352]: adjust time server 129.6.15.28 offset -0.102709 sec 17 Feb 13:25:01 ntpdate[5636]: adjust time server 129.6.15.28 offset -0.183842 sec So I haven't really fixed the root cause of the problem. And I'm betting that I'll be getting banned from the NTP server for my frequent time requests... Any ideas? -SeanO |
|
| Author: | kmkittre [ Sat Feb 17, 2007 3:43 pm ] |
| Post subject: | |
Just a guess, but the battery on your motherboard may be bad. |
|
| Author: | seano [ Mon Feb 19, 2007 9:59 pm ] |
| Post subject: | |
BIOS setup "PC Health Status" indicates that the battery is okay. Any other ideas? Thanks! -SeanO |
|
| Author: | abrendel [ Sun Mar 25, 2007 9:26 am ] |
| Post subject: | |
I've got the same exact problem here with the same motherboard (Dragon 1.1 spec). I wonder if Human has seen this with any of the other Dragon 1.0 and 1.1 spec'd machines? Any one else have ideas? I'm going to be changing my battery to test that, but it's a fairly new system. The battery shouldn't be dead yet. |
|
| Author: | Martian [ Sun Mar 25, 2007 11:50 am ] |
| Post subject: | |
I had an issue with severe clock drift at one point and it was solved by disabling the "spread spectrum" features in the BIOS. As weird as it sounds that is what worked for me and several others. Try Googleing "clock drift spread spectrum" and I think you will find more info. Martian |
|
| Author: | ceenvee703 [ Sun Mar 25, 2007 1:08 pm ] |
| Post subject: | |
Martian: I think he mentioned he disabled all the spread spectrum BIOS settings. I have the time problem every now and then; I don't think that amount of time adjustment is too much for a system that has the problem. I know there is a way to get ntpdate to select from a pool of ntp servers, rather than the same one each time. Just to triple-check, after you reconfigured lilo.conf, you reran lilo, right? And is your ntpd running? It is supposed to keep track of small differences and compensate. |
|
| Author: | thornsoft [ Sun Mar 25, 2007 3:19 pm ] |
| Post subject: | |
FWIW, I see it too, with my non-platinum version of this board (near dragon 1.0). My ntpd has died a couple times, and without it running, I'm several minutes fast by the end of the day. It gains something crazy like a minute an hour. The 6:00 news records fine, 8:00 shows seem a little "off", and then we miss several minutes of the 10:00 show. Usually Desperate Houswives or "Grey's". SORRY DEAR!! |
|
| Author: | seano [ Mon Mar 26, 2007 10:52 am ] |
| Post subject: | |
Hi All, I've rebooted a number of times... so I don't think I've missed restarting lilo. Everything spread spectrum has been disabled in bios. I'm glad I'm not the only one with the problem. I don't *think* I had this problem with the version prior to R5E50. The ntpdate in the cron every 5 minutes works - but is just a hack. I'd love to get to the bottom of this. ntpd couldn't keep up with the drift - only the cron'd ntpdate seems to work. And thornsoft: It's the "SORRY DEAR" factor that lead to the hack... -Sean |
|
| Author: | ceenvee703 [ Mon Mar 26, 2007 11:33 am ] |
| Post subject: | |
seano wrote: I've rebooted a number of times... so I don't think I've missed restarting lilo.
I'm no Linux guru, so I don't know if this is right, but my understanding is that after you modify lilo.conf you manually have to run lilo to make the changes "stick." I don't think rebooting is sufficient... |
|
| Author: | thornsoft [ Mon Mar 26, 2007 11:34 am ] |
| Post subject: | |
seano wrote: And thornsoft:
It's the "SORRY DEAR" factor that lead to the hack... LOL! |
|
| Author: | seano [ Mon Mar 26, 2007 11:44 am ] |
| Post subject: | |
ceenvee703 wrote: seano wrote: I've rebooted a number of times... so I don't think I've missed restarting lilo. I'm no Linux guru, so I don't know if this is right, but my understanding is that after you modify lilo.conf you manually have to run lilo to make the changes "stick." I don't think rebooting is sufficient... Bwaa! I'll try this tonight - thanks. -Sean |
|
| Author: | OwL [ Mon Apr 02, 2007 6:58 am ] |
| Post subject: | |
Try this: 1. add append="noapic" on the kernel image in lilo.conf 2. run: lilo 3. reboot you could also just press left ctrl at bootup, and add noapic on the boot parameters manually, to test it out... shen there is no need to run lilo or edit the config... it is wise to test it manually before inserting it permanently.. and test that out too see if you still get the clock-scew problem... you will probably just get too many calls to the timer interrupt, causing theses problems.... there is another kernel flag called "timerhack" I think.... could also try that... Regards // R |
|
| Author: | seano [ Mon Apr 02, 2007 10:39 am ] |
| Post subject: | |
I'm pretty sure that actually running lilo fixed my problem. I still have my cron job running - but looking at the logs I don't see any more "steps" - which are the big time jumps. I'm going to leave the cron for a while longer before I revert to ntpd. -SeanO |
|
| Author: | Mephi [ Fri Aug 31, 2007 11:00 am ] |
| Post subject: | |
I have the same issue (and a similar motherboard: MSI K8N Neo4-F), and I've tried these steps and it's not working. Does anyone know if there is anything else I can look at? |
|
| Author: | Mephi [ Sun Sep 02, 2007 12:50 pm ] |
| Post subject: | |
ok, after digging deeper, I think I've solved it. I found this link: http://ubuntuforums.org/showthread.php?t=53094 Which gave some other lilo.conf options to work through. the endof my append line now reads: Code: noapic acpi=noirq
I'm going to leave my ntpdate line running too, but I'm hopeful that this is it... |
|
| Page 1 of 2 | All times are UTC - 6 hours |
| Powered by phpBB® Forum Software © phpBB Group http://www.phpbb.com/ |
|