Getting error %STR-F-FATINTERR, fatal internal error

[Jeetu Singh]

Hi,

please help me to solve this issue as i am getting error during executing the below command:

ADVISE PERFORMANCE GRAPH/NODE_NAME=GAX058/TYPE=(CPU_UTILIZATION, -
MEMORY_UTILIZATION, -
COMPUTE_QUE,DISKS, -
MEMORY_UTIL, -
FAULTS, -
PROCESSES, -
TOP_HSC_THRUPUT,-
TOP_IO_VOLUMES,-
TOP_READ_VOLUMES,-
TOP_WRITE_VOLUMES,-
TOP_SPLITIO_VOLUMES,-
TOP_IOSIZE_VOLUMES,-
TOP_RESPONSE_TIME_VOLUMES,-
TOP_THRUPUT_VOLUMES) -
/FORMAT=POSTSCRIPT=(X_POINTS=336,CHAR=COLOR) -
/BEGIN=26-JUN-2010:00:00 -
/END=23-JUL-2010:23:59 -
/OUTPUT=PSPA$ROOT:[SYSTEMS.GAX058.COLOR.2010-06-26]
%STR-F-FATINTERR, fatal internal error


Waiting for your help.

[Phil.Howell]

does it also fail with a smaller list of types?

[Shriniketan Bhagwat]

Hi,

Welcome to OpenVMS ITRC forum.

From the online help it looks like the failure of an internal consistency check from Run-Time Library. Please follow the user action specified in the online help.

FATINTERR, fatal internal error

Facility: STR, String Manipulation Facility

Explanation: An explicit or implicit call to the OpenVMS Run-Time Library resulted in the failure of an internal consistency check.

User Action: Rerun your program. Collect as much information as possible, including the system traceback listing, and contact HP Customer Support.

Regards,
Ketan

[Phil.Howell]

In the olden days types were defined as
/TYPE=
([NO]graph_type[=([NO]STACK,Y_AXIS_MAXIMUM=n
,TITLE=string)],...
,ALL_GRAPHS[=([NO]STACK,THRESHOLD=m,Y_AXIS_
MAXIMUM=n)]
,CUSTOM=(see below ``/TYPE=CUSTOM''))

for more (old) info see
http://odl.sysworks.biz/disk$axpdocmar961/decw$book/d341aa24.p162.decw$book#1298

[Hoff]

With all due respect to the pointer to this being a bug in the RTL (and that error message text could certainly use an update), the OpenVMS RTLs do see extensive use throughout many applications and across disparate environments. Which means that while there can be bugs in VMS and its RTLs, Occam's Razor first points to bugs in the application code.

That this is more likely to be a bug within the CA PSPA / ADVISE / Performance Advisor tool.

Check for updates to whatever version of the tool is in use here, and (as Mr Howell indicates) make sure your command syntax is correct and complete, and (if there is a data file involved here) that your data file is consistent and uncorrupted; try different data files.

Then check with CA.

Even if this does prove to be evidence of an RTL bug, the CA application folks are the ones with the source code and the reproducer, and CA would need to work this with HP support. This after having a look at the application code around the call that failed, and then a look "upstream" at the heap and at the stack, as it's reasonably likely that there's been a calling error or a run-time corruption here.

[Hein van den Heuvel]

Forget about the runtime error suggestion. That's just a silly diversion.
Focus on DECps which is in trouble, probably due to a bug but that's probably just how it is going to stay.
So apply basic trouble shooting techniques.

1) did it ever work
2) what changed since since it stopped working
3) if you reduce the selected date range does it start working?
4) if you reduce the selected types does it start working? try just 1 first, then try half the list for a binary search.

5) does it fail right-away or does it make some progress?

6) running out of some (PGLFIL) quota?
if it fails slowly, watch it with SHOW PROC/CONT ... hit the Q key for quotas.

7) last but not least OpenVMS version? product version?

Good luck!
Hein

[abrsvc]

As others have stated, the problem is likely to be in the data or command line. I have also seen an error message generated by a call to LIB$signal that is incorrect. In otherwords, a status value that equates to the STR error is in fact generated by DECps NOT by an actual STR$ routine.

If this can be reproduced, than CA will need to see where the error occurs. I'm willing to bet that this error is a bad value passed to an erorr routine rather than a true runtime library problem.

Dan

[John McL]

Have you closely checked the syntax of the command, especially any limits to the number of types that can be specified?

Which version of VMS? Which version of ... is it PolyCenter Performance Advisor?

As someone else asked above, have you tried running the command with a much shorter number of types?

[John Gillings]

1) Commercial product
2) Syntactically correct command line (assumption)
3) Fatal error

Ergo, it's a bug.

Please report it to the manufacturer

You can mess with making the command line shorter if you wish, but that's really just doing the work of the support engineer.

This is not something that anyone without access to the source code will be able to help with.

[Fekko Stubbe]

I tried the same command on my system and it worked without a problem. But I do not have a month of data available, only too weeks. My version of the performance advisor is 2.4
Could it be that you run out of memory (pgflquota to low?). Does is work with a smaller interval?

Fekko Stubbe