Cycle SMAT with hcicyclesavemsgs

[Sandy McQuay]

Cloverleaf 5.2, AIX 5.2

I have a thread that will not cyclesave with hcicyclesavemsgs.  I can cycle it from the gui and from the command line but not from the script that is scheduled in cron.  Every other thread in the site will cycle, except this one.  

I’ve tried renaming the file then stopping/starting the process, but no dice.  I have not stopped the thread, opened SMAT, closed SMAT, then restarted the thread.  The “Save outbound messages” is selected in thread configuration.  There are no other thread names similar to this one.

Here’s how we’re set up

Cron job:

59 23 * * * /usr/integrator/scripts/cycle_sbmc_adt

cycle_sbmc_adt script

---

#!/usr/bin/ksh

setroot /qdx/qdx5.2/integrator

setsite sbmc_adt

#  This will cycle messages consistently with the night job.

#  No data will be lost EVER if this script is used.

/usr/bin/ksh -c ‘eval `/qdx/qdx5.2/integrator/bin/hcisetenv -root ksh /qdx/qdx5.

2/integrator/sbmc_adt`;hcicyclesavemsgs -d -o 120 -f /usr/integrator/CycleLog_sb

mc_adt’ >/tmp/cycle_sbmc_adt.log 2>&1

---

Here’s an excerpt from the log file:

Processing

   Root: /qdx/qdx5.2/integrator, site: sbmc_adt, proc: sbmc_adt1, thread: emcare_adt

       20080611.00:00:03 No saved IB msg file for this thread

       20080611.00:00:04 No saved OB msg file for this thread

I’ve talked to Quovadx (or Healthvision) support, but we haven’t been able to resolve this issue.  Any help would be greatly appreciated.

Topic

[Tom Rioux]

The entry from the log file mentions the process name as: sbmc_adt1

The screen shot you provided has the process name as: sbmctest_adt1

I’m not sure where the script is pulling the process name from but you may a good place to start.

Thanks…Tom Rioux

[Sandy McQuay]

My bad  😳 ; here’s the production thread screenshot (I didn’t make any changes, I swear!) 😀

[Jim Kosloskey]

Sandy,

You have probably already checked for this but make sure no other thread in the same site uses the same SMAT file name.

email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.

[Alan James]

…and, that is an outbound thread, right?

[Sandy McQuay]

Yeah it’s outbound – but I just noticed I did not have the Outbound Only box checked . . . I checked it and saved the config and will bounce the process shortly.  I’ll see what happens tonight when the cycle script runs.

Jim – the save file name is unique.  I’ve even changed it a couple of times to try to get it to cycle.  Thanks.

[Gary Atkinson]

Check your log for the same thread being picked up twice by the perl script.  I had an issue where I had two threads with similar names, so one of the threads was getting picked up twice.  Also, I make sure my thread names are less than 15 characters, so hciconnstatus displays all the thread name.

[Sandy McQuay]

Checking the Outbound Only box didn’t help.  The thread name is unique and under 15 characters.  Here are all the thread names in the site:

Process            Connection    

sbmc_adt2       adt_dump2

sbmc_adt1       mercuryMD_adt

sbmc_adt1       wellsoft

sbmc_adt1       hh_adt

sbmc_adt1       sunquest_adt

sbmc_adt1       lab_orr

sbmc_adt1       adt1

sbmc_adt1       rals_adt

sbmc_adt1       medrite_adt

sbmc_adt1       sr_health

sbmc_adt2       radiology

sbmc_adt1       watchchild

sbmc_adt2       adt2

sbmc_adtIB      adt_IB

sbmc_adt1       sbbh_mpi

sbmc_adt2       inv_2_sci

sbmc_adt2       clinic_ar_adt

sbmc_adt2       rad_med

sbmc_adtIB      adt1_LH

sbmc_adt1       misys_adt

sbmc_adt2       clinic_neaa_adt

sbmc_adt2       ipath

sbmc_adt2       bed_track

sbmc_adt2       mos

sbmc_adt2       OPC_Quantim

sbmc_adt1       drg_qtm_epf

sbmc_adt1       bal_epf

sbmc_adt2       muse_adt

sbmc_adt1       pharmacy_orders

sbmc_adt1       adt_epf

sbmc_adt1       medmined

sbmc_adt2       aria_adt

sbmc_adt1       emcare_adt

sbmc_adt2       pyxis

sbmc_adt2       pyxis_rx

sbmc_adt1       adt_dump

sbmc_adtIB      adt2_LH

sbmc_adt2       quantim

sbmc_adt2       pacs_adt

Any other thoughts or ideas?  Thanks for all the suggestions.

Sandy

[Gary Atkinson]

When you run the perl script also output hciconnstatus.  What is the status of the thread in question?

[Michael Hertel]

Start the threads then check the process log to see if the save failed.

It’s possible the messages are not being saved in the first place.

[Sandy McQuay]

Michael – The SMAT files continue to grow.  I have to manually cycle the output (when I remember to do it), so I assume the messages are being saved.

Gary – Are you saying output hciconnstatus from within the hcicyclesavemsgs perl script?  I’m not too good with Perl.  Or could I do it before/after the perl script runs.

[Rob Abbott]

The issue here is that hcicyclesavemsgs doesn’t think that OB saving is turned on for this particular thread.  It gets this information from hciconndump.

What is the output from hciconndump -v emcare_adt (run this after a setroot to your production site)

Rob Abbott
Cloverleaf Emeritus

[Sandy McQuay]

Here’s what I get:

HCI root is /qdx/qdx5.2/integrator

HCI site is sbmc_adt

/home/hci>hciconndump -v emcare_adt

Dump of connection(s): emcare_adt

---

 Process: sbmc_adt1            Connection: emcare_adt                    

 EO Proc:

EO Config:

  EO Msg:

  Groups:

Data Type: frl                               Options: AUTO

RecoverDb: Yes               Hold: No       SendOnly: Yes

Disk msgs: No

Save Msgs:  in: No   Inbound File:

          out: Yes Outbound File: emcare_adt_OB

Comm Type: fileset

       Mode: local-tps

IB    style: nl               Scan time: 30        Read time: 5  

IB max msgs: 1

IB      dir:

IB  dir TPS:

IB  del TPS:

OB    style: single           OB append: No  

OB      dir: /usr/integrator/emcare/prod/adt/out

OB     file: emcare.adt

---

[Gary Atkinson]

I’ll fish again  😆

What does hciconnstatus say for this thread when you run the perl script? Is it up?

In sub ckconnStatus it has this line:

Code:

if (/^S+s+(S+)s+ups+S+.*$/)

[Rob Abbott]

Here is the logic with some debugging added.  Sandy save this to your Cloverleaf system as “cycle-test”.  

Code:

$conn = “emcare_adt”;
$proc = “sbmc_adt1”;
local($filepath) = “”;
$rootpath = $ENV{’HCIROOT’};
$site = $ENV{’HCISITE’};

$procpath = $rootpath . “/” . $site . “/exec/processes/” . $proc;
print “procpath = ‘” . $procpath . “‘n”;

  open(CONNDUMP,”hciconndump -v $conn | grep ” out: ” |”)
     || warn “Cannot access hciconndump”;
  while () {
     chop;
     #; check OB flag
     $isOutbound = 0;
     if (m/s+out:s+Yess+(S+)s+File: (.*)/) {
        ($cycletype,$type,$stype,$saveFile) = (’out’,’Outbound’,’OB’,$2);
print “match foundn”;
        local($filepath) = $procpath . “/” . $saveFile . “.idx”;
print $cycletype . “, ” . $type . “, ” . $stype . “, ” . $saveFile . “n”;
print “filepath = ‘$filepath’n”;
        #; cycle only if file is not empty
        if ( -s $filepath ) {
print “file exists and is not emptyn”;
           $isOutbound = 1;
           &manageSaveFiles(0);# before cycle
           &cycleSavemsgs if -s $filepath;
           &manageSaveFiles(1);# after cycle
        } else {
print “file does not exist or is emptyn”;
       }
     }
}

print “conndump completen”;
print “isOutbound = ” . $isOutbound;

Run perl cycle-test after a setroot and post the output.

Rob Abbott
Cloverleaf Emeritus

[Sandy McQuay]

Here’s the output from Rob’s code:

HCI root is /qdx/qdx5.2/integrator

HCI site is sbmc_adt

/usr/integrator/scripts>perl cycle-test

procpath = ‘/qdx/qdx5.2/integrator/sbmc_adt/exec/processes/sbmc_adt1’

match found

out, Outbound, OB, emcare_adt_OB

filepath = ‘/qdx/qdx5.2/integrator/sbmc_adt/exec/processes/sbmc_adt1/emcare_adt_OB .idx’file does not exist or is empty

conndump complete

isOutbound = 0

Looks like a space in the file name.  I opened Netconfig and clicked in the filename field under Outbound on the Thread tab.  Sure enough there was a space after the B (i.e. “emcare_adt_OB “).  I deleted the extra space, saved Netconfig, bounced the thread and ran cycle-test again.  Here’s what I got this time:

procpath = ‘/qdx/qdx5.2/integrator/sbmc_adt/exec/processes/sbmc_adt1’

match found

out, Outbound, OB, emcare_adt_OB

filepath = ‘/qdx/qdx5.2/integrator/sbmc_adt/exec/processes/sbmc_adt1/emcare_adt_OB.idx’

file exists and is not empty

Undefined subroutine &main::manageSaveFiles called at cycle-test line 27, line 1.

So, maybe the space was causing all this  :roll:.  I’ll let it cycle save normally tonight and update everyone tomorrow.  Thanks for all the suggestions (so far 🙂 ).

[Sandy McQuay]

Woo Hoo!

[Author]

Replies