CRONTAB

[Stewart]

I have a script that calls another script to transfer files to SFTP. It works correctly if I run manual but not with crontab.  It will not call the commands from this program using crontab.  Any idea what would cause this?

#!/opt/cloverleaf/cis19.1/integrator/tcl/bin/expect -f

Topic

[Stewart]

adding email notifications

[David Barr]

You probably don’t have LD_LIBRARY_PATH and TCL_LIBRARY set to the right values. I think these are usually set up by the “setroot” command which is called from your login scripts. You may need to make a shell script that calls your expect script, and the shell script can set the environment before calling the script.

[Keith McLeod]

I have a new install on Red Hat Enterprise Linux release 8.5 (Ootpa).  I have worked through several issues regarding LD_LIBRARY_PATH by modifying the environment in .profile.local.end for commands such as scp, sftp, rsync and now mail.  I received the following error prior to the modification: mail: symbol lookup error: mail: undefined symbol: SSLv3_client_method, version OPENSSL_1_1_0.

I added to .profile.local.end and this seemed to resolve from the command line.:

function mail {
if [[ -n “$HCIROOT” && -n “$CL_INSTALL_DIR” ]]; then
clinstalldir=$CL_INSTALL_DIR
hcisite=$HCISITE
hciroot=$HCIROOT
setroot -clear
/usr/bin/mail “$@”
export CL_INSTALL_DIR=$clinstalldir
setroot $hciroot $hcisite
else
/usr/bin/mail “$@”
fi
}

This works for using mail at the command line, however inside many of my notification tcl scripts I have code like:

set sm [open “| mail -s \”${client} – [string toupper $HciSite]: [string toupper $thdName] Not showing a Connection for [expr $downtime / 60] minutes.\” -c ie_mail,${systemName}_mail -r $fromAddress $mailAddress” w]
puts $sm $assign_to
puts $sm $priority
puts $sm $contact
puts $sm ${client_id}
puts $sm $description

close $sm

This brings me back to the symbol lookup error.  Any thoughts on this???

 

[Peter Heggie]

What Dave said. Cron does not have a normal user context associated with it, so many of the environment variables (not just Cloverleaf’s variables, but also ‘normal’ user variables) are not set.

Depending on whether you are issuing cloverleaf commands or just running UNIX commands that need the HCI user’s context, you could use one of these two styles:

1. with no Cloverleaf commands:

05 03 * * * setroot /hci/cis19.1/integrator; . cronpath ; mediquant_to_epsi >> $CRONLOGS/mediquant_to_epsi.log 2>&1

• where “mediquant_to_epsi” is the application script

2. with Cloverleaf commands:

30 00 * * * setroot /hci/cis19.1/integrator; /usr/bin/ksh -c ‘eval /hci/cis19.1/integrator/sbin/hcisetenv -root ksh /hci/cis19.1/integrator soabatprd;. cronpath ; soarian_to_pcp_edchart_fax_stop >> $CRONLOGS/soarian_to_pcp_edchart_fax_stop.log 2>&1’

• where “soarian_to_pcp_edchart_fax_stop” is the application script

fyi “. cronpath” will execute a proc called cronpath which sets additional environment variables (like $CRONLOGS – which resolves to a directory name). Using the “dot space name” syntax means execute that named script and apply the results to the current environment instead of a child environment.

Peter Heggie

[Stewart]

This is the crontab setting.  I tried setting the environment variables from env output before the script1 was called but no luck.  I also tried setting these in /etc/crontab.

0,10,20,30,40,50 * * * *  /home/hci/scripts/dinh_inc_SCRIPT2.sh >> /home/hci/scripts/logs/ftp_dinh_inc.log

This is the section of script2 which calls script1 which is responsible for uploading the files.  I’ve been able to write to a log file from the called script before and after the commands but the commands are ignored.  So the script was getting called but the commands aren’t.

sleep 3
$home/scripts/dinh_inc_SCRIPT1.sh >> $logfile
FTP_ERROR_MSG=”Failure”

#!/opt/cloverleaf/cis19.1/integrator/tcl/bin/expect -f

set timeout 120
log_file “/home/hci/scripts/logs/ahmedgresftp.log”
send_log — “####-Starting SFTP script – [exec date] \n”
send_log — “Today’s date:[exec date] \n”
send_log — “Downloading files through sftp \n”
spawn sftp name@address

 

[Robert Kersemakers]

What we used to do with cron was first switch the user to ‘hci’ before executing the cron-script. This way all environment variables were set correctly:

00 12 * * 6 su – hci -c “/cldata/scripts/weekly_housekeeping_cloverleaf” > /dev/null 2>&1

 

Nowadays we do something different by first calling a different script, for ex.:

55 23 * * * /home/hci/sethcienv daily_housekeeping_cloverleaf > /cldata/scripts/daily_housekeeping_cloverleaf.log 2>&1

The script ‘sethcenv’ looks like this:

if [[ -d /prod/cloverleaf/sites62 ]]
then
. /etc/profile
. /home/hci/.profile
echo “+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++”
$*
fi
exit 0

It first checks whether the server the cron is running is the active server (we have an active/passive high-availability setup). If this is the case the environments are set after which the required scripts are run.

Hope this helps.

Zuyderland Medisch Centrum; Heerlen/Sittard; The Netherlands

[Stewart]

Thanks for all the suggestions.  I got this resolved by reviewing the mail messages in /var/spool/mail/hci.  First error stated it couldn’t locate the expect file, once this was resolved by placing SHELL and PATH at the top of crontab -e it generated another error.  The second error was that it couldn’t locate the shared libraries so I placed the TCL_LIBRARY and LD_LIBRARY output from env below PATH in crontab -e.  I added the following to the top of the crontab.

SHELL=/bin/sh
PATH=/opt/cloverleaf/cis19.1/integrator/prod_adt/bin:/opt/cloverleaf/cis19.1/integrator/prod_adt/scripts:/opt/cloverleaf/cis19.1/integrator/prod_master/bin:/opt/cloverleaf/cis19.1/integrator/prod_master/scripts:/opt/cloverleaf/cis19.1/integrator/bin:/opt/cloverleaf/cis19.1/integrator/contrib:/opt/cloverleaf/cis19.1/integrator/sbin:/opt/cloverleaf/cis19.1/integrator/dbms/bin:/opt/cloverleaf/cis19.1/integrator/tcl/bin:/opt/cloverleaf/cis19.1/integrator/clgui/bin:/opt/cloverleaf/cis19.1/integrator/clgui/java/bin:/opt/cloverleaf/cis19.1/integrator/usercmds:/usr/bin:/etc:/usr/sbin:/usr/ucb:/home/hci/bin:/usr/bin/X11:/sbin:.:/usr/local/bin:/usr/local/scripts:/bin:/home/hci/bin
TCL_LIBRARY=/opt/cloverleaf/cis19.1/integrator/tcl/lib/tcl8.6
LD_LIBRARY_PATH=/opt/cloverleaf/cis19.1/integrator/clgui/java/lib/amd64:/opt/cloverleaf/cis19.1/integrator/clgui/java/lib/amd64/server:/opt/cloverleaf/cis19.1/integrator/lib:/opt/cloverleaf/cis19.1/integrator/bin:/opt/cloverleaf/cis19.1/integrator/tcl/lib:/opt/cloverleaf/cis19.1/integrator/dbms/lib:/usr/lib:/lib:/shlib:/usr/shlib:/opt/cloverleaf/cis19.1/integrator/lib/perl5/5.22.0/x86_64-linux/CORE:/opt/cloverleaf/cis6.2/lib/

[Rob Lindsey]

This is how we do it with our 20.1 environment using RHEL

18,48 * * * * source /home/hci/.profile ; /hci/cloverleaf/scripts/export_interface_info_tables_to_web_page_v3.tcl >> /info/logs/export_interface_info_tables_to_web_page_v3_$(date +\%Y\%m\%d).log 2>&1

The source command pulls in the profile which makes sure that all of the Libraries are set correctly for our environment.

Rob

[Author]

Replies