Bug 75558
Summary: | ntpd incorrectly requires an option for -x parameter | ||||||
---|---|---|---|---|---|---|---|
Product: | Red Hat Enterprise Linux 2.1 | Reporter: | Need Real Name <robert_macaulay> | ||||
Component: | ntp | Assignee: | Harald Hoyer <harald> | ||||
Status: | CLOSED ERRATA | QA Contact: | Brian Brock <bbrock> | ||||
Severity: | medium | Docs Contact: | |||||
Priority: | medium | ||||||
Version: | 2.1 | CC: | joshua, lwoodman, tao | ||||
Target Milestone: | --- | ||||||
Target Release: | --- | ||||||
Hardware: | i386 | ||||||
OS: | Linux | ||||||
Whiteboard: | |||||||
Fixed In Version: | Doc Type: | Bug Fix | |||||
Doc Text: | Story Points: | --- | |||||
Clone Of: | Environment: | ||||||
Last Closed: | 2003-02-14 21:02:09 UTC | Type: | --- | ||||
Regression: | --- | Mount Type: | --- | ||||
Documentation: | --- | CRM: | |||||
Verified Versions: | Category: | --- | |||||
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||
Cloudforms Team: | --- | Target Upstream Version: | |||||
Embargoed: | |||||||
Attachments: |
|
Description
Need Real Name
2002-10-09 21:33:43 UTC
good point... dooh.. there will be an erratum Is this available in an errata yet? Larry Woodman waiting in the QA queue... sorry Will the anticipated errata also keep ntp from slightly skewing the time when using the -x option? The recommended -g parameter in addition to the step tickers file to correct the drift has no effect. The clock continues to slew out of sync. Ntp produced the same results. Once ntp begins to slew the clock, it will never slew the time back into sync and ntp continues to adjust the time futher out of sync. Regardless of the time source, ntp works correctly if it is set to step time. ntp fails miserably if it slews the time. Also, ntp-4.1.1a-8.AS21.1 still has the same problem as the prior versions. We are seeing many customers with this problem. Please, please fix. hmm, ntp uses adjtime, maybe the error is there?? Ulrich, Jakub? Any known issues? Also, should ntp use ntp_adjtime?? All adjtime is doing is converting the time back and forth from and to struct timeval format and then it calls ntp_adjtime. ntp_adjtime takes the time as an integer in 5sec. It should make much of a difference to call ntp_adjtime directly. you may want to try ftp://people.redhat.com/ntp/4.1.1a-11.AS21.1 This version uses ntp_adjtime vs. __adjtime in the old version. Please give it the -x slew test.. ftp://people.redhat.com/harald/ntp/4.1.1a-11.AS21.1 that is... sorry ftp://people.redhat.com/harald/ntp/4.1.0b-2.AS21.1/ is with minimal changes... erratum is in the queue Created attachment 89416 [details]
fix ntpd -x skew bug
This patch is to fix ntpd -x skewing bug.
Apply to "/usr/src/linux-2.4/kernel/time.c".
This problem causes kernel bug. The do_adjtimex() function in /usr/src/linux-2.4/kernel/time.c is wrong implementation. When the time difference of a server and a client becames 0.512 or more, ntpd stop skewing. Since a clock is set by "step" with a difference for 0.128 seconds while not using -x options, this is happend only while using -x option. This implementation is very strange in comparison with other OS's, such as FreeBSD, Solaris and HP-UX. Then, I corrected Linux kernel to reference for the source code of FreeBSD. (see http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/kern/kern_ntptime.c? rev=1.49&content-type=text/x-cvsweb-markup) Try my patch. An errata has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on the solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHBA-2002-230.html |