Mills@udel.edu (01/03/91)
Folks, I can't help Nathan, since I am creatively allthumbs with respect to Unix gurudom. Could somebody help? In response, I would be glad to receive and archive reports of successful ports. It would be a good idea anyway and help offload those of you who have been so helpful in the past. Dave ----- Forwarded message # 1: Received: from louie.udel.edu by huey.udel.edu id aa24030; 2 Jan 91 15:05 EST Received: from osprey.ncsc.org by louie.udel.edu id aa09321; 2 Jan 91 15:03 EST Received: by osprey.ncsc.org (5.59/tas-concert/4-5-90) id AA22369; Wed, 2 Jan 91 15:03:18 EST Date: Wed, 2 Jan 91 15:03:18 EST From: "Nathan H. Hillery" <hillery@ncsc.org> Message-Id: <9101022003.AA22369@osprey.ncsc.org> To: mills@udel.edu Subject: xntpd on convex/sgi/3090? Sorry to bother you, but I was wondering if you knew of anyone who has gotten xntpd to work on Convex's, SGI's, or IBM3090's. I have gotten the standard distribution to compile (with a minor change since Convex's UNIX includes getopt() in libc), but it doesn't seem to adjust the time at all. I cannot get "tickadj -s" to run, since there is no "dosynctodr" symbol. I assume that this is the problem. The system has an adjtime() system call and the variables "tick" and "tickadj". I have run into the same problem on our IBM3090 running AIX. It doesn't use any of the BSD-style names for the kernel time-keeping variables. I haven't looked as much at our Silicon Graphics, but know (and have heard from others) that standard xntpd won't work. It appears to have the kernel variable "tick", but nothing else. It has something called "timetrim" which may be like tickadj. Thanks for any information, Nate. .............................................................................. . Nathan H. Hillery North Carolina Supercomputing Center (NCSC) . . hillery@ncsc.org P.O. Box 12889 . . Voice: 919/ 248-1106 3021 Cornwallis Road . . FAX: 919/ 248-1101 Research Triangle Park, NC 27709-2889 . .............................................................................. ----- End of forwarded messages
iglesias@draco.acs.uci.edu (Mike Iglesias) (01/03/91)
In article <9101021547.aa24657@huey.udel.edu> Mills@udel.edu writes: >Folks, > >I can't help Nathan, since I am creatively allthumbs with respect >to Unix gurudom. Could somebody help? In response, I would be glad >to receive and archive reports of successful ports. It would be a >good idea anyway and help offload those of you who have been so >helpful in the past. I have ntp (not xntp) running on our Convex C240. It's been a while since I built it, but I don't remember doing anything special to make it work. Our Convex was losing 4 seconds a day before we started running ntp on it. Mike Iglesias University of California, Irvine Internet: iglesias@draco.acs.uci.edu BITNET: iglesias@uci uucp: ...!ucbvax!ucivax!iglesias
jones@HERMES.CHPC.UTEXAS.EDU (01/03/91)
If you are running an older convex c120`s, c130`s, and older c200's then you have a problem. The older machines loose clock interrupts when they talk to the spu because of a kernel "printf". The two most common cases that a convex looses time because of a printf the settimeofday function and nfs errors - timeouts. The settimeofday function loose about 200 ms because of the messages the is sent to the spu. Which is bad since this is larger then 128 ms which means that xntp and ntp will always stay unsynchronized. You can get a patch to noop the printf in the settimeofday function. If you have a lot of nfs mounts you will also see about a 200 ms glitch in the time every time you have a nfs error. If you have a real new convex, we have a convex C220 wide body, then you may not see the above problems. If you don't have a lot of nfs errors I would suggest getting the settimeofday patch from convex. You don't have to run tickadj with the "-s" option, its not needed. Just run tickadj with the "-A" option or the "-a" option to set the tickadj to a smaller value then 1 ms. Bill Jones
whaley@elbo.ucs.ubc.ca (Paul Whaley) (01/03/91)
In article <9101021547.aa24657@huey.udel.edu>, Mills@udel.edu writes:... |> to Unix gurudom. Could somebody help? In response, I would be glad... |> From: "Nathan H. Hillery" <hillery@ncsc.org> |> Subject: xntpd on convex/sgi/3090? |> Sorry to bother you, but I was wondering if you knew of anyone |> who has gotten xntpd to work on Convex's, SGI's, or IBM3090's. ... |> I have run into the same problem on our IBM3090 running AIX. It doesn't |> use any of the BSD-style names for the kernel time-keeping variables. We have tried to run ntpd on AIX/370, with no success. I tracked it down to problems in the adjtime system call, which seems to be so broken that there is no way to get ntp or xntp to run. Calls to adjtime to make small changes (microseconds) in the time-of-day result in a time that is too small by seconds. After a few hours of running ntpd, the clock is so far off that ntpd gives up. This is on AIX/370 with updates PTF U401122 installed (alias AIX/370 01.02.0300). We reported the problem to IBM at the end of October, but have had no reply yet. Until it's fixed, we are using the ntpdate program from the xntp distribution. The ntpd program does seem to run fine on AIX PS/2, after adjustment for the errors in conversions BOTH WAYS between float and unsigned long, and a fix for the little-endian placement of int:8. |> I haven't looked as much at our Silicon Graphics, but know (and have |> heard from others) that standard xntpd won't work. ... We are running ntpd on our Silicon Graphics 4D25s with IRIX 3.3.0, without any apparent problem. It has to be compiled for the BSD signal semantics: (in the Makefile: INCPATH= -I/usr/include/bsd LIBS= -lbsd -lmld CC=cc -g FEATURES= -DDEBUG DEFINES= -D_BSD_SIGNALS ) We haven't tried xntpd on the SGI's or IBM/3090. -- Paul Whaley Internet: whaley@ucs.ubc.ca Academic Operating Systems UUCP: ...!ubc-cs!mtsg.ubc.ca!Paul_Whaley University Computing Services BITNET: USERWHAL@UBCMTSG U. of British Columbia Phone: (604) 228-3976