T4 and Restart

[James T Horn]

T4$COLLECT .... Version: X-28 IDENT 'X-28'21-May-2008

When I have to restart the system, the T4$COLLECT job is not working through the reboot.

16:32:26) $ Submit/Notify/Restart/Keep/NoPrint/Log=UCSTMP:[T4$SYS.DATA]T4$COLLECT_DEC6 -
DISK$ALPHAES40:[VMS$COMMON.T4$SYS]T4$COLLECT.COM/Queue=T4$BATCH -

/Parameter=("NOW/"+P1_1, "MON/", "T4$BATCH", "EWB0:", "", "60", "UCSTMP:[T4$SYS.DATA]", "9999,9999,9999,0,0,0,N,0,,N,N")
%DCL-W-NOPAREN, value improperly delimited - supply parenthesis
\NOW/\
(16:32:26) $ EndIf

[Hein van den Heuvel]

Looks like someone fat-fingered a script.

What are those slash "/" doing there in the
strings "NOW/" and "MON/" ?

Check timestamps and versions for the files?

Hein

[John Gillings]

James,

I believe T4 relies on too much customisation in the scripts. Stuff that should probably be done with site specific callouts, but then I don't expect any significant changes or improvements in the future.

Fix the errors Hein has pointed out. Even then, I've found that interrupting T4 doesn't always do the right thing to cleanup the files and restart the collectors.

Most likely another case of OpenVMS being a victim of its own reliability! Since failures are rare, the recovery code isn't always completely debugged for all cases.

On our site, I've extracted the T4 POST_PROCESS section from T4$COLLECT as a separate procedure. There's a regular job that checks the T4 data area for evidence of an interrupted collections, and performs the post process. We also have a process monitor that will restart a missing collector (since we do a fair bit of destructive testing, we tend to have more than the normal frequency of systems crashed or ripped from the cluster).

Modifying T4 procedures is non-trivial. You'll need a mechanism for merging and distributing your changes into any updates from T4.

[Ranjan Sarangi · Accepted solution]

> 16:32:26) $ Submit/Notify/Restart/Keep/NoPrint/Log=UCSTMP:[T4$SYS.DATA]T4$COLLECT_DEC6 -
> DISK$ALPHAES40:[VMS$COMMON.T4$SYS]T4$COLLECT.COM/Queue=T4$BATCH -

> /Parameter=("NOW/"+P1_1, "MON/", "T4$BATCH", "EWB0:", "", "60", "UCSTMP:[T4$SYS.DATA]", "9999,9999,9999,0,0,0,N,0,,N,N")
> %DCL-W-NOPAREN, value improperly delimited - supply parenthesis
> \NOW/\

While submitting the t4 collector it fails since all the 8 parameters passed into it are
not in correct format. T4$COLLECT.COM is expecting a end time after NOW/ within quote (as a parameter).
I guess someone changed the original T4$COLLECT.COM script. Please check the script for
any modification. If you can post the T4$COLLECT.COM script which you are using now, we
can have a look on it.

[James T Horn]

Here is the T4$COLLECT.COM file.

[Art Wiens]

Someone definately edited your T4$COLLECT.COM procedure. Mine has the same version listed but a few sections are different ... I'ld say an attempt was made to introduce some sort of monthly arrangement (where your's is X.X):

File SYS$SYSDEVICE:[SCGTS]X.X;1
$ month = f$element(f$cvtime("yesterday",,"month"),"/","/JAN/FEB/MAR/APR/MAY/JUN/JUL/AUG/SEP/OCT/NOV/DEC/")
$ File_name = "SYS$MANAGER:NELL" + f$cvtime("yesterday",,"day") + month + f$cvtime("yesterday",,"year") + ".DAT"
$ Set Command T4$Sys:T4$Aclg
******
File _$1$DGA403:[T4.T4$SYS]T4$COLLECT.COM;1
$ Set Command T4$Sys:T4$Aclg

Time to find the author, or debug 1353 lines of DCL goodness yourself.

I attached the original if you give up ;-)

Art

[Hein van den Heuvel]

The "/" is "NOW/" is the intended value. No worry there. That line looks clean.

My guess now is that someone re-defined the SUBMIT command for you.

Check out $ SHOW SYMB SUBM*

Hein.

[Art Wiens]

There are a few "issues" with the edited com:

$ @dcl_check x.x

Checking file SYS$SYSDEVICE:[SCGTS]X.X;


Procedure contains: 1353 total lines
723 code lines (including 28 lines w/ comments)
81 additional continuation lines
0 lines w/i $DECK/$EOD pairs
426 comment only lines
123 blank lines

LINE CODE --DIAGNOSTIC MESSAGE--
398 PRQ probable error using single-quote (')
406 PSQ possible error using single-quote (') in quoted string
495 LNR label "START_PROC" not referenced (warning)
499 PRQ probable error using single-quote (')
527 LNR label "MANAGE_STORAGE" not referenced (warning)
615 PRQ probable error using single-quote (')
619 TLS target label "'BATCH$RESTART'" provided by symbol substitution (warning)
622 LNR label "START_DATA_COLLECTION" not referenced (warning)
631 UMP unmatched parentheses
632 UMP unmatched parentheses
654 LNF label "" not found
668 PRQ probable error using single-quote (')
679 PRQ probable error using single-quote (')
690 PRQ probable error using single-quote (')
698 WCT wrong constant type for comparison
700 PRQ probable error using single-quote (')
716 PRQ probable error using single-quote (')
733 WCT wrong constant type for comparison
735 PRQ probable error using single-quote (')
746 PRQ probable error using single-quote (')
763 PRQ probable error using single-quote (')
771 PRQ probable error using single-quote (')
795 PRQ probable error using single-quote (')
800 PRQ probable error using single-quote (')
842 PRQ probable error using single-quote (')
843 PRQ probable error using single-quote (')
853 PRQ probable error using single-quote (')
915 PRQ probable error using single-quote (')
936 PRQ probable error using single-quote (')
966 PRQ probable error using single-quote (')
977 PRQ probable error using single-quote (')
1005 WCT wrong constant type for comparison
1010 PRQ probable error using single-quote (')
1010 LNF label "" not found
1026 WCT wrong constant type for comparison
1033 WCT wrong constant type for comparison
1040 WCT wrong constant type for comparison
1048 WCT wrong constant type for comparison
1055 WCT wrong constant type for comparison
1161 LNR label "EXIT_AND_RESUBMIT" not referenced (warning)
1233 PRQ probable error using single-quote (')
1240 LNR label "GII_NO_FILE" not referenced (warning)
1245 PRQ probable error using single-quote (')
1286 PRQ probable error using single-quote (')
1299 PRQ probable error using single-quote (')
1347 PRQ probable error using single-quote (')
1349 PRQ probable error using single-quote (')
-*- END OF LISTING -*- 12-OCT-2009 11:44:29.22

Do a diff against nl: to get the line numbers.

Cheers,
Art

[Art Wiens]

Actually the "issues" don't appear to be related to the edits. Amazing what you can get away with and still have DCL work! ;-)

Art