Can I set a timeout to cp?

[Muthukumar_5]

Eric,

You are setting TIMEOUT after cp operation getting completed based upon cp command return status.

If you want to set TIMEOUT for cp operation while occuring, then you have to start it background and check TIME for not reaching more than TIMEOUT variable.

You can upgrade your check by checking connectivity with remote server (ping).

which operation you are requiring putting TIMEOUT for cp command or after completion of cp command?

hth.

[Eric Antunes]

Hi Muthukumar,

You are right!

I added this:

...

backup_started=0

while [ $exit_code -eq 1 ]
do
if [ $backup_started -eq 0 ]
then
backup_started=1
echo "\nStarting copy..."
cp -Rp $SOURCE_DIR/* $DESTINATION_DIR
exit_code=$?
fi

...

And I'm checking the results...


Best Regards,

Eric Antunes

[Muthukumar_5]

You can try this script.

#!/bin/ksh
WAIT_TIME=5
TIMEOUT=300
TIMER=0

ORIGEM=/disc1/oradata/TST
DESTINO=/_disc2/
exit_code=1
FLAG=0

while [ $exit_code -eq 1 & $FLAG -eq 1 ]
do
echo "waiting for backup...\n"
sleep ${WAIT_TIME}
TIMER=`expr $TIMER + ${WAIT_TIME}`

cp -Rp $ORIGEM/* $DESTINO
exit_code=$?
if [[ $? -eq 0 ]]
then
let FLAG=1
fi

if [ $TIMER -ge $TIMEOUT ]
then
echo "\nWARNING: backup NOT EXECUTED!\n";
exit_code=8;
let FLAG=1
fi
done

echo "\nSUCCESS! Backup executed!!\n"

hth.

[Muthukumar_5]

sorry. Use while loop as,
while [ $exit_code -eq 1 && $FLAG -eq 1 ]

hth.

[Eric Antunes]

The problem is: when the NFS server is up everything runs ok but, when it is down (actual situation), it stops at the cp instruction...

Maybe, starting cp in background mode:

cp -R $SOURCE_DIR/* $DESTINATION_DIR &

But I still need to check if cp ended ok before TIMEOUT!! How can I pass the return code of a command started in background mode??

Thanks,

Eric Antunes

[Eric Antunes]

Will this work, anylysing the cp.log and, simultaneously, counting?

cp -R $SOURCE_DIR/* $DESTINATION_DIR > /disc2/.../cp.log
2>&1

[Sandman!]

> The problem is: when the NFS server is up everything runs ok but, when it is
> down(actual situation), it stops at the cp instruction.

that's exactly why you need to first ping the nfs server && then kick-off the copy.

ping && cp -R

The purpose here is that the entire command line works or fails as a unit. Only iff the ping succeeds does the copy kickoff and if the ping fails the copy never gets executed.

see script attached...

enjoy!!!

[A. Clay Stephenson]

Just because ping works (or for that matter doesn't work) is really no indication of the status of NFS. Moreover, the return status of ping is far from reliable. It will reurn non-zero for host not found but 0 if it can ing but with errors.

If this were me I would do something like this:

df -F nfs /desiredmountpoint

If that returns a zero status then you can assume that NFS is available. Of course, you need a timeout mechanism for this as well.

This would be my approach:

TDIR=${TMPDIR:-/var/tmp}

timeoutcmd.pl -t 20 "df -F nfs /desiredmountpoint" > /dev/null 2>/dev/null
STAT=${?}
if [[ ${STAT} -eq 0 ]] # df ok
then
timeoutcmd.pl -t 5400 "cp -p file1 file2 /desiredmountpoint" > /dev/null 2>/dev/null
STAT=${?}
if [[ ${STAT} -ne 0 ]]
then
echo "Cp failed; status ${STAT}" >&2
fi
else
echo "NFS unavailable; status ${STAT}" >&2
fi
exit ${STAT}


Note: This is not the same Perl script posted earlier:

[Eric Antunes]

Hi Clay,

I'm testing your perl script now. Is there any good manual about perl?

Hi Sandman,

The problem is that the command "ping && cp -R " never ends when the server is down (actual situation to test the script): I have to kill the ping processes...

Best Regards,

Eric Antunes

[Eric Antunes]

Clay,

With the NFS server down I get "NFS unavailable; status 127".


With the NFS server up, I still get "NFS unavailable; status 127"...

Thanks,

Eric Antunes

[Eric Antunes]

Ok,

From the command line, I get:

interpreter "/usr/bin/perl" not found

[Muthukumar_5]

can you find perl utility as,

find / -name "perl"

which perl

hth.

[Eric Antunes]

I found one perl interpreter in /usr/contrib/bin/perl...

[Muthukumar_5]

When you are going to ping then use finite count as,

ping -n 1

ping -n 1 1>/dev/null 2>&1
if [[ $? -eq 0 ]]
then
action;
else
echo 'server failed'
fi

hth.

[Eric Antunes]

Now (with /usr/contrib/bin/perl), I get:

"NFS unavailable; status 255"

[Muthukumar_5]

Is # /usr/contrib/bin/perl -e 'print "hello world!\n"' printing hello world in shell. Then change /usr/bin/perl to /usr/contrib/bin/perl in olympian's script and then script will start to work.

hth.

[Muthukumar_5]

Eric,

Did you modify clay's script?

Are you using this?
STAT=${?}
if [[ ${STAT} -ne 0 ]]
then
echo "Cp failed; status ${STAT}" >&2
else
echo "NFS unavailable; status ${STAT}" >&2
fi
exit ${STAT}

please post that script to help you out.

Before testing any script try it command line with single commands.

hth.

[Eric Antunes]

I'm using this script:

WAIT_TIME=30
TIMEOUT=540
TIMER=0

SOURCE_DIR=/disc2/app/oravis/admin/TST/bdump/

DESTINATION_DIR=/NFS_SERVER_disc2/bdump_PROD

TEMP_DIR=${TMPDIR:-/var/tmp}

timeoutcp_clay2.pl -t 20 "df -F nfs ${DESTINATION_DIR}" > /dev/null 2>/dev/null
STAT=${?}
if [[ ${STAT} -eq 0 ]] # df ok
then
timeoutcp_clay2.pl -t 540 "cp ${SOURCE_DIR}* ${DESTINATION_DIR}" > /dev/
null 2>/dev/null
STAT=${?}
if [[ ${STAT} -ne 0 ]]
then
echo "Cp failed; status ${STAT}" >&2
else
echo "Cp completed; status ${STAT}" >&2

fi
else
echo "NFS unavailable; status ${STAT}" >&2
fi
exit ${STAT}


And I found why. Executing the following from the command line:

$export DESTINATION_DIR=/hp440_disc2/bdump_PROD
$x=timeoutcp_clay2.pl -t 20 "df -F nfs ${DESTINATION_DIR}" > /dev/null 2>/dev/null; echo $x

I get nothing (null) into x...

Thanks,

Eric Antunes

[Eric Antunes]

About "ping -n 1 1>/dev/null 2>&1", it return always "Server OK", even if the server is down...

[Eric Antunes]

Of course I've changed first line of timeoutcp_clay2.pl from #!/usr/bin/perl -w to #!/usr/contrib/bin/perl -w

[Muthukumar_5]

Try this script.

#!/bin/ksh
WAIT_TIME=30
TIMEOUT=540
TIMER=0

SOURCE_DIR=/disc2/app/oravis/admin/TST/bdump/

DESTINATION_DIR=/NFS_SERVER_disc2/bdump_PROD

TEMP_DIR=${TMPDIR:-/var/tmp}

if [[ -f ./timeoutcp_clay2.pl ]]
then
echo "Script is available"
else
echo "exiting"
exit 1
fi

./timeoutcp_clay2.pl -t 20 "df -F nfs ${DESTINATION_DIR}" > /dev/null 2>/dev/null
STAT=${?}

if [[ ${STAT} -eq 0 ]] # df ok
then
./timeoutcp_clay2.pl -t 540 "cp ${SOURCE_DIR}* ${DESTINATION_DIR}" > /dev/null 2>/dev/null
STAT=${?}
if [[ ${STAT} -ne 0 ]]
then
echo "Cp failed; status ${STAT}" >&2
else
echo "Cp completed; status ${STAT}" >&2
fi
else
echo "NFS unavailable; status ${STAT}" >&2
fi
exit ${STAT}


#

Put ./timeoutcp_clay2.pl in the current path with this script. It will work.

hth.

[Muthukumar_5]

Nop. When you do ping as,

ping -n 1 1>/dev/null 2>&1

will return 0 or 1 code based upon hostname machine up.

#!/bin/ksh
echo "Enter remote machine name to check"
read host
ping $host -n 1 1>/dev/null 2>&1
if [[ $? -eq 0 ]]
then
echo "$host is running"
else
echo "$host is down"
fi

#################

hth.

[A. Clay Stephenson]

The Perl in /usr/contrib is very, very old (Perl 4.x); you need to install a newer version Perl 5.x; nobody uses Perl 4 except for some HP scripts which expects Perl 4.

Perl can be found on the Applications CD's or dowmloaded from www.perl.org/CPAN. In either case it's an easy swinstall.
-------------------------------------------

However, there was a bug in the the timeoutcmd.pl script. Unlike the C equivalent pclose() function, Perl's close() function returns 0 on a failure. I needed to reverse the login and also take care of the case where the exit status of a command was exactly divisible by 256 because exit() only returns the low-order 8 bits.

This version should do exactly what you want.

NOTE: the data is send to stdout so if you did a command ls "ls -l /etc" using timeoutcmd.pl, the ls is sent to stdout.

[Eric Antunes]

Ok,

It seems to be working now with perl 5.x!

With the NFS server down, I get:

"Script is available
sh[55]: 11128 Alarm call
NFS unavailable; status 142"


With the NFS server up, I get:

"Script is available
Cp completed; status 0"

Many thanks,

Eric Antunes

[A. Clay Stephenson]

Ooops, I'm an idiot (or the attachment failed) (or both) but the improved timeoutcmd.pl script attachement from my last posting is missing.

Here's the improved version with the close() function returning the correct logic and ensuring that the low-order 8 bits reurned by exit() will be 0 if okay and non-zero otherwise.