Hi all... I''ve just got my hands on a Solaris 10 box for the first time. I''ve heard wonderous things about dtrace, so the first command I type is dtrace -l which disappointingly does nothing at all. It just hangs, and neither Ctrl-D, Ctrl-C or kill -9 from another shell will do anything to kill it. The box is: SunOS t2000 5.10 Generic_118822-25 sun4v sparc SUNW,Sun-Fire-T200 top reports it using around 3% cpu, which corresponds to one hardware thread on the cpu i guess. These are the last few lines from truss -fo /tmp/dtrace.out dtrace -l 1323: ioctl(3, DTRACEIOC_CONF, 0x10010D340) = 0 1323: openat(-3041965, "/system/object", O_RDONLY|O_NDELAY|O_LARGEFILE) = 4 1323: fcntl(4, F_SETFD, 0x00000001) = 0 1323: fstat(4, 0xFFFFFFFF7FFFF640) = 0 1323: getdents(4, 0xFFFFFFFF7F404000, 8192) = 4016 And that''s where it hangs... no output from dtrace to stdout, nor to the truss output file. Any ideas? thanks, Richard This message posted from opensolaris.org
> Hi all... > > I''ve just got my hands on a Solaris 10 box for the first time. I''ve heard wonde > rous things about dtrace, so the first command I type is dtrace -l which disappo > intingly does nothing at all. It just hangs, and neither Ctrl-D, Ctrl-C or kill > -9 from another shell will do anything to kill it. > > The box is: > SunOS t2000 5.10 Generic_118822-25 sun4v sparc SUNW,Sun-Fire-T200 > > top reports it using around 3% cpu, which corresponds to one hardware thread on > the cpu i guess. > > These are the last few lines from truss -fo /tmp/dtrace.out dtrace -l > > 1323: ioctl(3, DTRACEIOC_CONF, 0x10010D340) = 0 > 1323: openat(-3041965, "/system/object", O_RDONLY|O_NDELAY|O_LARGEFILE) = 4 > 1323: fcntl(4, F_SETFD, 0x00000001) = 0 > 1323: fstat(4, 0xFFFFFFFF7FFFF640) = 0 > 1323: getdents(4, 0xFFFFFFFF7F404000, 8192) = 4016 > > And that''s where it hangs... no output from dtrace to stdout, nor to the truss o > utput file. > > Any ideas?Hmm: not good. Is it possible for you to take a crash dump to help us debug this? (this will involve your machine taking a reboot lap) If that''s not impossible for you, and since you seem to still be able to log into the system, become root and do a "reboot -d" in your shell. (make sure one of the hung dtrace processes is running and hung when you do this) After reboot, the crash dump will be saved for you in /var/crash/`uname -n`/ Once you have it, e-mail me offline and we''ll get it transferred to us. If it isn''t possible for you to do that to your machine, let us know and we can walk you through how to debug this live. -Mike -- Mike Shapiro, Solaris Kernel Development. blogs.sun.com/mws/
Early builds of Solaris 10 HW2 shipped with beta units of Sun Fire T2000 had problems with DTrace. Please install Solaris 10 12/05 (aka unofficially Solaris 10 Update 1) which is now available for download. Regards, Jignesh Richard Gray wrote:>Hi all... > >I''ve just got my hands on a Solaris 10 box for the first time. I''ve heard wonderous things about dtrace, so the first command I type is dtrace -l which disappointingly does nothing at all. It just hangs, and neither Ctrl-D, Ctrl-C or kill -9 from another shell will do anything to kill it. > >The box is: >SunOS t2000 5.10 Generic_118822-25 sun4v sparc SUNW,Sun-Fire-T200 > >top reports it using around 3% cpu, which corresponds to one hardware thread on the cpu i guess. > >These are the last few lines from truss -fo /tmp/dtrace.out dtrace -l > >1323: ioctl(3, DTRACEIOC_CONF, 0x10010D340) = 0 >1323: openat(-3041965, "/system/object", O_RDONLY|O_NDELAY|O_LARGEFILE) = 4 >1323: fcntl(4, F_SETFD, 0x00000001) = 0 >1323: fstat(4, 0xFFFFFFFF7FFFF640) = 0 >1323: getdents(4, 0xFFFFFFFF7F404000, 8192) = 4016 > >And that''s where it hangs... no output from dtrace to stdout, nor to the truss output file. > >Any ideas? > >thanks, >Richard >This message posted from opensolaris.org >_______________________________________________ >dtrace-discuss mailing list >dtrace-discuss at opensolaris.org > >
> Early builds of Solaris 10 HW2 shipped with beta units of Sun Fire T2000 > had problems with DTrace. > Please install Solaris 10 12/05 (aka unofficially Solaris 10 Update 1) > which is now available for download. > > Regards, > JigneshPlease do not do that (yet). And Jignesh, while we appreciate trying to help customers solve problems, kernel hangs are not to be taken lightly. We don''t ask people to install patches when we haven''t root-caused the problem yet. I have the crash dump from Richard, I''m analyzing it, and when we have the actual problem diagnosed, we''ll suggest an appropriate course of action. In this case, given what I''ve seen so far, Richard''s problem isn''t going to be solved by upgrading to 12/05: this is a different problem which is unrelated to DTrace per se. -Mike -- Mike Shapiro, Solaris Kernel Development. blogs.sun.com/mws/
Mike,> Please do not do that (yet). And Jignesh, while we appreciate trying to help > customers solve problems, kernel hangs are not to be taken lightly. We don''t > ask people to install patches when we haven''t root-caused the problem yet.Agreed completely - the days of "just install the latest patchset will solve (all) your problem(s)" are, and should be long gone.> I have the crash dump from Richard, I''m analyzing it, and when we have the > actual problem diagnosed, we''ll suggest an appropriate course of action. > In this case, given what I''ve seen so far, Richard''s problem isn''t going > to be solved by upgrading to 12/05: this is a different problem which is > unrelated to DTrace per se.Would it be at all possible if you could publish here the potted details of how your investigation proceeded. I realise that this is/could be time-consuming, but I am sure that it would be appreciated by all. Thanks, regards, and Happy New Year... Sean.
Hi All, It looks like this problem might be hardware related. We''ve noticed errors in /var/adm/messages after rebooting to clear out some hung processes. The sys admins here are engaging Sun through a more formal channel. I''ll let you know how we get on there. Richard On 12/31/05, Michael Shapiro <mws at sun.com> wrote:> > > Early builds of Solaris 10 HW2 shipped with beta units of Sun Fire T2000 > > had problems with DTrace. > > Please install Solaris 10 12/05 (aka unofficially Solaris 10 Update 1) > > which is now available for download. > > > > Regards, > > Jignesh > > Please do not do that (yet). And Jignesh, while we appreciate trying to help > customers solve problems, kernel hangs are not to be taken lightly. We don''t > ask people to install patches when we haven''t root-caused the problem yet. > > I have the crash dump from Richard, I''m analyzing it, and when we have the > actual problem diagnosed, we''ll suggest an appropriate course of action. > In this case, given what I''ve seen so far, Richard''s problem isn''t going > to be solved by upgrading to 12/05: this is a different problem which is > unrelated to DTrace per se. > > -Mike > > -- > Mike Shapiro, Solaris Kernel Development. blogs.sun.com/mws/ >