[Orca-users] Problems displaying some orca graphs

Charles Dennett charles.dennett at kodak.com
Sat Feb 26 04:59:32 PST 2000


From: Charles Dennett <charles.dennett at kodak.com>

Yesterday I sent a message to this list about some problems I was
having displaying some of the graphs produced by orca. (I did not
receive a copy of the message.  I assume the list is set up not to
send copies back to the sender.)

To summarize that, running orca 0.25 and orcallator 1.23 (was
orcallator 1.20 until I noticed a new version on Blair's ftp
site last night - also grabbed corresponding orcallator.cfg
file) and SE Toolkit 3.1-preFCS.  All this on some Sun
Ultras (1 and 2) under Solaris 2.6

Most of the graphs display fine.  Some never come up.  The web server
on the main machine is Netscape Enterprise Server 3.63.  One strange
behavior I noticed is that the httpd process aborts when it goes
after these missing graphs.  If I select the Daily graphs for one
of the servers and then try clicking on one of the broken links,
I get a "Page contained no data" error in my browser. I used
Solaris' truss command to trace the httpd process as I did this.  The
results are below.  The httpd process aborts when I tried to 
grab this image.

Now, this is just a wild guess, but these files have very long filenames.
I made a link with a very short name and pointed it to one of the
files that causes the problem.  Then I called that in my browser.  The
graphs appead fine.  Is it possible that the netscape web server 
is choking on these very long file names?  The only graphs that
consistantly do this are the Disk Run Percent, Disk Space Percent Usage
and Disk Inode Percent Usage.  I believe the other occasional missing
graphs are due to the browser making several connections to the web server
at one time and one of those connections is trying to get the problem
graph which aborts the server so all graphs being retrieved at that
time fail.  I can always click on those other missing ones and get the
graphs.  I can never get the 3 graphs mentioned above.

Anyone seen this before?  Is there a Solaris patch?  Any other kind of patch?

Here's the truss output I mentioned:

read(13, " G E T   / o r c a / o _".., 2048)    = 605
fcntl(13, F_GETFD, 0x00000000)                  = 0
fcntl(13, F_SETFD, 0x00000001)                  = 0
getsockopt(13, 6, 1, 0xEE4C3A94, 0xEE4C3A98)    = 0
brk(0x0015A020)                                 = 0
brk(0x00162020)                                 = 0
stat("/web/htdocs/orca/o_toolchest_disk_runp_c0t6d0,__disk_runp_c0t0d0,__disk_runp_c1t2d0,__disk
_runp_c1t1d0,__disk_runp_c2t11d0,__disk_runp_c2t13d0,__disk_runp_c0t1d0,__disk_runp_c2t2d0,__dis
k_runp_c2t10d0,__disk_runp_c2t4d0,__disk_-_zu8Sn9cuNaXED2RZx9SZQ.html", 0x0015C888) = 0
open("/web/htdocs/.nsconfig", O_RDONLY)         Err#2 ENOENT
open("/web/htdocs/orca/.nsconfig", O_RDONLY)    Err#2 ENOENT
open("/dev/zero", O_RDWR)                       = 15
mmap(0x00000000, 147456, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE|MAP_NORESERVE, 15, 0) = 0xE
DAB0000
close(15)                                       = 0
mprotect(0xEDAB0000, 8192, PROT_NONE)           = 0
lwp_create(0xEE4C32D8, 0x00C0, 0xEDAD3E14)      = 12
lwp_continue(12)                                = 0
lwp_create(0x00000000, 0, 0x00000000)           = 0
lwp_self()                                      = 12
lwp_schedctl(SC_STATE, -1, 0xEDAD3D7C)          = 0
lwp_mutex_unlock(0xEF369DC8)                    = 0
lwp_mutex_lock(0xEF369DC8)                      = 0
lwp_self()                                      = 12
    Incurred fault #6, FLTBOUNDS  %pc = 0xEF0A45CC
      siginfo: SIGSEGV SEGV_MAPERR addr=0x78656432
    Received signal #11, SIGSEGV [caught]
      siginfo: SIGSEGV SEGV_MAPERR addr=0x78656432
sigprocmask(SIG_SETMASK, 0xEF3668E8, 0x00000000) = 0
sigaction(SIGSEGV, 0xEDAD3548, 0x00000000)      = 0
setcontext(0xEDAD33A8)
    Incurred fault #6, FLTBOUNDS  %pc = 0xEF0A45CC
      siginfo: SIGSEGV SEGV_MAPERR addr=0x78656432
    Received signal #11, SIGSEGV [default]
      siginfo: SIGSEGV SEGV_MAPERR addr=0x78656432
        *** process killed ***
--
Charles Dennett
Eastman Kodak Company, Rochester, NY
Sitemaster - http://www.kodak.com

------------------------------------------------------------------------
Get one email address for all your friends!  
Start a free email group on eGroups!
http://click.egroups.com/1/1881/5/_/576712/_/951570179/
------------------------------------------------------------------------




More information about the Orca-users mailing list