Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Mailing Lists -> Oracle-L -> Re: Is this a well known ORACLE bug? Max uptime < 1 year
Hi,
This is what I found out. The problem is with all versions of Oracle on Solaris 2.6. ( Fixed in 8.1.7 ) 64bit versions of Solaris ( 2.7 and above ) should not have this problem. Anyway this problem has only been reported on 2.6 and the patch is only available for that platform and version.
Oracle uses the times system call and it returns a long ( structure clock_t ). On 32bit Solaris the maximum value can only be 2^31 but on 64bit Solaris it can be 2^63. I also checked on HP and it returns an unsigned integer ( both 32 and 64 bit ) and this problem should occur only after 497 days of uptime.
Regards,
Denny
"Adams, Matthew (GEA, 088130)" wrote:
>
> Is this bug specific to Solaris?
> I ask because on the bug header it list platform as solaris, but
> it also lists affected platforms as generic.
>
> I tried to ask that question on one of the metalink forums, but
> keep getting an error.
>
> > -----Original Message-----
> > From: Paul van Dijken [mailto:paul.vandijken_at_sema.nl]
> > Sent: Wednesday, August 23, 2000 8:48 AM
> > To: Multiple recipients of list ORACLE-L
> > Subject: RE: Is this a well known ORACLE bug? Max uptime < 1 year
> >
> >
> > From another Oracle list:
> >
> > >
> > > It's a legitimate bug (1084273/1227119) that is Solaris-specific.
> > > There is a patch available for it outside of 8.1.7, however. The
> > > patch is available on oracle-ftp.oracle.com under
> > > /server/outgoing/SUN_SOLARIS2/BUG1227119
> > >
> > > The analyst said that the patch is for the 32-bit version, and is
> > > checking to see if the patch needs to be applied to the 64-bit
> > > version as well.
> > >
> >
> > Since most systems have booted on january, 1, 2000, You can
> > expect the first
> > effects on September 4, 2000 around noon.
> >
> > Paul
> >
> > -----Original Message-----
> > Sent: 23 August 2000 11:02
> > To: Multiple recipients of list ORACLE-L
> >
> >
> > Dear listers,
> >
> > From another system we learned that ORACLE 8.0.5
> > should not be started if the unix (Sun ?) has
> > been running for a very long time.
> >
> > Instead you should boot the unix first.
> >
> > uptime is the magic command giving you the
> > number of days since the last boot.
> >
> > I tried to figure out what goes wrong and
> > my theory is this: Unix keeps and ORACLE gets
> > the value of uptime as 1/100 of a second.
> > Max (signed int) = 2^31 corresponds to 248,55
> > days (as 1/100 seconds). Then unsigned int
> > is interpreted by ORACLE as negative =>
> >
> > 2698: sigprocmask(SIG_UNBLOCK, 0xEFFFE420, 0x00000000) = 0
> > 2698: setcontext(0xEFFFE4F0)
> > 2690: semop(524288, 0xEFFFE808, 1) (sleeping...)
> > 2694: Received signal #14, SIGALRM, in semop() [caught]
> > 2694: semop(524288, 0xEFFFE808, 1)
> > Err#91 ERESTART
> > 2694: sigprocmask(SIG_BLOCK, 0xEFFFE420, 0x00000000) = 0
> > 2694: times(0xEFFFE3B0) =
> > -1944987590
> > 2694: setitimer(ITIMER_REAL, 0xEFFFE3B0, 0x00000000) = 0
> > 2694: sigprocmask(SIG_UNBLOCK, 0xEFFFE420, 0x00000000) = 0
> > 2694: setcontext(0xEFFFE4F0)
> > 2706: Received signal #14, SIGALRM, in semop() [caught]
> > 2706: semop(524288, 0xEFFFE808, 1)
> > Err#91 ERESTART
> > 2706: sigprocmask(SIG_BLOCK, 0xEFFFE420, 0x00000000) = 0
> > 2706: times(0xEFFFE3B0) =
> > -1944987544
> >
> > We were told that this has been fixed in 8.1.7
> > but can that be true ?
> >
> > Do you know this ? Is this a problem in 8.0.5.2 or
> > 8.0.6 or 8.1.6 ?
> >
> > There is a bell ringing somewhere in the back of my head someone on
> > this list wrote something about rebooting at least ones a
> > year. Is this
> > related?
> >
> > Best regards / Met vriendelijke groeten / Ystävällisin terveisin
> >
> > Eric Lansu
-- Denny Koovakattu http://www.vitalsol.com/ http://www.oneco.net/ _________________________________________________________Received on Fri Aug 25 2000 - 13:36:50 CDT