Hi, If this isn''t the correct place to post this question, please excuse me then. After running iotop (dtrace) on our Fire E6900 for about 5sec the complete machine crashed. No console reply, empty logfiles ... no errors found. The machine was under a heavy IO stress. Did someone had this issue before ? Thanks, Wesley. This message posted from opensolaris.org
is there a crash dump in /var/crash/<hostname>? if so, do this (as root): # cd /var/crash/hostname # mdb -k N (where N is the highest number for unix.N and vmcore.N) you''ll get a prompt ">" denoting mdb is waiting for input:> $Cand the output - supplied to this list - may help.> [press ^D to exit]HTH Michael On Mon, 29 May 2006 07:24:48 -0700 (PDT) wesley hof <wesley.hof at scarlet.biz> wrote:> Hi, > > If this isn''t the correct place to post this question, please excuse > me then. > > After running iotop (dtrace) on our Fire E6900 for about 5sec the > complete machine crashed. No console reply, empty logfiles ... no > errors found. The machine was under a heavy IO stress. Did someone > had this issue before ? > > Thanks, > Wesley. > > > This message posted from opensolaris.org > _______________________________________________ > dtrace-discuss mailing list > dtrace-discuss at opensolaris.org-- Michael Schuster (+49 89) 46008-2974 / x62974 visit the online support center: http://www.sun.com/osc/ Recursion, n.: see ''Recursion''
G''Day Wesley, On Mon, 29 May 2006, wesley hof wrote:> Hi, > > If this isn''t the correct place to post this question, please excuse me > then. > > After running iotop (dtrace) on our Fire E6900 for about 5sec the > complete machine crashed. No console reply, empty logfiles ... no errors > found. The machine was under a heavy IO stress. Did someone had this > issue before ?Darn - I''ve never seen that before (I wrote iotop). I have heard of iotop throw out probe drop warnings, which is on my todo list to improve (what''s stopped me is that I don''t have access to a machine that does this - I''ll try harder on my test servers :). So, under extreme load I wouldn''t be too suprised if iotop slowed a server down for a while; but DTrace always sorts itself out after a minute or two. No console reply sounds dire - and sad that you can''t do a "sync". I can''t immediatly think of why DTrace would interfere with that.. thanks for the email, Brendan
Morning,>If this isn''t the correct place to post this question, please excuse me then.>After running iotop (dtrace) on our Fire E6900 for about 5sec the complete machine crashed. No console reply, empty logfiles ... no errors found. The machine was under a >heavy IO stress. Did someone had this issue before ?This might have nothing to do with DTrace or iotop(DTT) itself. It is enough to fire dtrace and trigger a bug in kernel and there you go... Hard to say. See the previous thread talk: http://www.opensolaris.org/jive/thread.jspa?threadID=7344&tstart=105 Are you using Solaris 10 GA, U1 ? Are you able to reproduce this bug ? Im testing iotop as well other DTT utilities but I have never seen a system crash. Rgds, Stefan
Hi, One extra note: The bug I mentioned seems to be related only to Niagara systems. http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6274126 http://www.opensolaris.org/jive/thread.jspa?threadID=7344&tstart=105 You might need to provide a stack trace of the kernel core to see whats going on there. stefan