scope and glance - resource allocation issue

[royalliege]

Hello,

There is a strange problem where I want to check resource utilization with glance but it directly creates a core file and OVO SCOPE process isn't running on the server although I have stopped and started it several times. It gives a bad allocation message where there is enough memory&swap space and parameter usage seems fine (i checked the forum and encountered http://community.hpe.com/t5/Languages-and-Scripting/aCC-runtime-error/m-p/4287919/highlight/true#M43448) Could it be a known bug or anything you would suggest to investigate in core file?

status.scope output

**** /opt/perf/bin/scopeux : 01/22/16 10:42:50 ****
scopeux/HP-UX 11.13.007 COLLECTOR BEGIN.
aCC runtime: Uncaught exception of type "std::bad_alloc".
aCC runtime: what(): bad allocation exception thrown (0xffffff30 bytes)
server02:/var/opt/perf#
server02:/var/opt/perf#
server02:/var/opt/perf#swapinfo -tam
Mb Mb Mb PCT START/ Mb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 4096 0 4096 0% 0 - 1 /dev/vg00/lvol2
dev 49152 0 49152 0% 0 - 1 /dev/vg00/lvswap01
reserve - 53248 -53248
memory 466899 171697 295202 37%
total 520147 224945 295202 43% - 0 -
server02:/var/opt/perf#
server02:/var/opt/perf#kcusage maxdsiz
Tunable Usage / Setting
=============================================
maxdsiz 638255104 / 1073741824
server02:/var/opt/perf#kcusage maxdsiz_64bit
Tunable Usage / Setting
=============================================
maxdsiz_64bit 1187708928 / 2147483648

[Dennis Handly]

>It gives a bad allocation message where there is enough memory & swap space and parameter usage seems fine

aCC runtime: Uncaught exception of type "std::bad_alloc".
aCC runtime: what(): bad allocation exception thrown (0xffffff30 bytes)

What is the value of maxdsiz_64bit?  You need to grow by 4 Gb.

 
maxdsiz_64bit XXXX / 2147483648

This is less than half the needed size.  (The proper command is kctune(1m)).

[royalliege]

I also have other HPUX 11.31 servers running glance with that setting of maxdsiz_64bit. How did you understand that it is less than required? 

Regards,

[royalliege]

**** /opt/perf/bin/scopeux : 01/25/16 15:39:07 ****
scopeux/HP-UX 11.13.007 COLLECTOR BEGIN.

So it started again; after killing midaemon process. mwa stop; mwa start action apparently didn't kill midaemon process. That was the solution, the cause is still unknown. It still may occur.

[Dennis Handly]

>How did you understand that it is less than required?

Because it failed?  :-)  Because (0xffffff30 bytes) is near 4 GB.

Of course I don't know why it needs so much.

[Dennis Handly]

>Because (0xffffff30 bytes) is near 4 GB.

Hmm, since scopeux is a 32 bit app, 0xffffff30 is really a negative number so likely to be a software bug.