Skip Menu |
 

This queue is for tickets about the Protocol-Modbus CPAN distribution.

Report information
The Basics
Id: 51251
Status: open
Worked: 30 min
Priority: 0/
Queue: Protocol-Modbus

People
Owner: cosimo [...] cpan.org
Requestors: gtp [...] lbl.gov
Cc:
AdminCc:

Bug Information
Severity: (no value)
Broken in: (no value)
Fixed in: (no value)



Subject: /Library/Perl/5.8.8/Protocol/Modbus/Transport/TCP.pm uninitialized value error
Date: Mon, 09 Nov 2009 13:57:51 -0800
To: bug-Protocol-Modbus [...] rt.cpan.org
From: gtp <gtp [...] lbl.gov>
Download (untitled) / with headers
text/plain 1.2k
My perl script, which is run one time every minute by root's crontab, returned one of these errors to me today. It's the first I've seen in months of operation. Use of uninitialized value in concatenation (.) or string at /Library/Perl/5.8.8/Protocol/Modbus/Transport/TCP.pm line 27. Can't connect to Modbus server on 128.32.117.40: at /Library/Perl/5.8.8/Protocol/Modbus/Transaction.pm line 63 It's possible that the error in TCP.pm was induced/caused/exacerbated by my modbus module which occasionally (a couple of times a day) barfs out hundreds or thousands or tens of thousands of unexpected/unsolicited TCP packets, or ICMP, or ARP packets. My guess is that this error path in TCP.pm is taken rarely, and I was just lucky enough to see an occurrence. I hope you find this helpful for cleaning up the code. I presume my cron script died when the error occurred and I didn't get my data from the I/O module. If you revise the package I'd appreciate an email so I can download a new version, assuming it will make my data collection more reliable... Thank you for your otherwise excellent resource. Gerald Przybylski for the Electronics shop of the Physics department of the University of California, Berkeley 510 642 3834
Download (untitled) / with headers
text/plain 104b
Can you please confirm that you are using the latest CPAN version (0.05)? perl is 5.8.8. What platform?
Subject: Re: [rt.cpan.org #51251] /Library/Perl/5.8.8/Protocol/Modbus/Transport/TCP.pm uninitialized value error
Date: Mon, 09 Nov 2009 15:00:24 -0800
To: bug-Protocol-Modbus [...] rt.cpan.org
From: gtp <gtp [...] lbl.gov>
Download (untitled) / with headers
text/plain 1.3k
Hi, Perl is 5.8.8 Protocol Modbus is 0.05; it was installed a couple of months ago, and the newest version is from late last year, I believe. The server is a Macintosh Mini running 10.5.8 of the operating system. The Modbus TCP device that gives me unexpected packets is the Advantech ADAM-6050 (which I would never buy again!!) The server and I/O module are on the same subnet, but on different ports of the ethernet 100-Base-T switch. They're in the berkeley.edu address space, not on a VLAN. I was running dumpcap at the time this error occurred, so you can look through the capture file with wireshark if you would find that useful. It happened at exactly 2:30pm today (20091109223011 GMT), according to the emailed report. The file is http://ln2.physics.berkeley.edu/log/ws/echo_00067_20091109054027 My reading of the file is that it occurred when the modbus/tcp I/O module was misbehaving. The script (attached) attempts first to ping the device, then opens the port (SYN), then reads the module, then closes (FIN). In this case the ICMP response never arrived, and the script attempted to connect several times. A screen snapshot is attached too. Jerry Cosimo Streppone via RT wrote: Show quoted text
> <URL: https://rt.cpan.org/Ticket/Display.html?id=51251 > > > Can you please confirm that you are using the latest CPAN version > (0.05)? perl is 5.8.8. What platform? >
Download test43.pl
text/x-perl 8.8k
#!/usr/bin/env perl # # Base on 'Modbus/TCP Server query' by Cosimo Feb 1st, 2007 # Issues a read coils request on an IP address / port # Here Protocol::Modbus is used only to build request # $Id: read_coils_simple.pl,v 1.2 2007/02/05 11:16:02 cosimo Exp $ # # UC Berkeley Physics -- Electronics shop -- gtp 3/19/09 # This script runs as a cron job every minute or longer. # This script creates a three files; a last-read file; a history file; a feature-state file. # See initialization below. # The feature-state file is delivered to the user's browser by a cgi script. # changes: Nov 4, 2009 migrate from basic read paradigm to general paradigm use strict; use warnings; use carp; use Protocol::Modbus; use Net::Domain; use Sys::Hostname; use Socket; my $netdebug = 0; if ( ( defined $ARGV[0] ) && ( $ARGV[0] =~ m/debug/ ) ) { $netdebug=1; } # Turn on the tcp debugging my $version = ' V1.0' ; my $modbusHost = q!echo.physics.berkeley.edu!; #my $ip = join(/./,unpack("C4",(gethostbyname($modbusHost))[4])); my $ip = '128.32.117.40'; # the IP of the ethernet input device my $trials = 5; my $datapath = "/var/www/Documents/LN2/snapshot/"; # Where non-cgi files are located #my $datapath = "/Users/electronicsshop/embedded/"; # Where non-cgi files are located my $logpath = "/var/log/" ; #my $logpath = "/Users/electronicsshop/embedded/" ; my $lndata = $datapath.'lndata'; # Formatted snapshot of the last device read my $lnhistory= $logpath.'lnhistory'; # Ultimately holds date-stamped changes of state. my $snapshot = $datapath.'last' ; # bit-pattern my $reportRoute = "1" ; my @features = ( "Sub-floor Temperature:", "Sub-floor Water Level state:", "Atmosphere (per ODM):", # future home of oxygen depletion monitor "Scale:", "Scale BP:", # bypass switch on interlock system state "Chain:", "Permit:", "VB-Permit:",# The input pin of the Nitromatic's controller. "START:", # Nitromatic "Start" button state "Station:", # Nitromatic Fill System "Filling" state "Station:", # Small Dewar Solenoid Valve Status "Power:"); my @featureStates =( "Normal#<font color=\"blue\">Subnormal. Clears within 15 min. unless there is a system fault.</font><font color=\"black\"></font>", "Normal#Flooded. Potential scale sensor damage from immursion.", "Safe#<font color=\"red\">DANGER!! Anoxic! UNSAFE TO ENTER WHILE ALARM SOUNDS!!</font><font color=\"black\"></font>", "Fill Permitted#Fill Disallowed", "Bypass Mode#Normal Mode", "Complete#Open Interlock", "Normal#Not Granted", "Permit#STOPPED", "DEPRESSED#Normal", "Filling#Not Filling", "Filling#Not Filling", "On#Off"); $| = 1; my $mylast = q!000000000000!; if ( open( RFH, "<".$snapshot ) ) { # Recall data from the preavious pass through the program, if it exists $mylast = <RFH>; close( RFH ); } my $timestamp = localtime(); # a time stamp very close to the actual read time. my $sysstatus = 0; if ($netdebug) { $sysstatus = `/var/www/CGI-Executables/dotcpdump start echo 128.32.117.40`; # print 'started dump at '.localtime().'status-'.$sysstatus."-\n"; } # the dotcpdump script runs tcpdump as a bsackground process, and captures the PID of it in a file so it can be killed later # The "stop" invocation reads the PID from a file, and issues a kill command # If the 'connect' succeeds, the tcp dump file can or should be deleted with the "unsave" invocation of the script # usage dotcpdump start|stop|unsave [filename [hostname|ip]] my $areyouthere = `/sbin/ping -qc1 -t1 $ip` ; # ping once to see if there's a path to the host. Produces a 2 packet ICMP transaction if ( $areyouthere =~ /100%\s+packet\+loss/) { if ($netdebug) { $sysstatus = system("/var/www/CGI-Executables/dotcpdump stop echo") ;} if ( $mylast =~ /path error/ ) { # Still out of service exit 0; # exiting cleanly prevents cron from sending emails; die provikes email reports } else { open( LOGFH, ">".$snapshot ) or die " can't open snapshot file after ping failure \n"; # The beginning of an outage print LOGFH "path error ".$timestamp.$version; close LOGFH; open( LOGFH, ">>".$lnhistory ) or die "can't open the log file after ping failure \n"; print LOGFH "Ping failed. No network path to I/O module at ".$timestamp; close LOGFH; die "Ping failed. No network path to $ip at $timestamp \n"; } } my $name = hostname(); # Make a randomizer to avoid collisions if multiple pollers my $randomizer = ((unpack("C4",(gethostbyname($name))[4]))[3])/16; # The MODBUS dialog begins here. my $modbus = Protocol::Modbus->new(driver=>'TCP'); my $req = $modbus->readInputsRequest( address => 0, quantity => 12, ); # Open a Socket to Device # use IO::Socket::INET; my $trial=$trials ; select(undef, undef, undef, $randomizer ); # the dotcpdump script runs tcpdump as a bsackground process, and captures the PID of it in a file so it can be killed later # The "stop" invocation reads the PID from a file, and issues a kill command # If the 'connect' succeeds, the tcp dump file can or should be deleted my $trs ; until ( defined( $trs = Protocol::Modbus::Transport->new( driver=>'TCP', address => $ip, PeerPort => 502, Timeout => 3, ))) # Even works with ADAM-6050 with Timeout=1... i.e. no connect errors. { if ( 0 < --$trial ){ if ( $reportRoute && ( $trials-$trial == 2 ) ) { # $route = system("/usr/local/bin/lft -SVV echo.physics.berkeley.edu") ; # Possible place for a diagnostic # LFT doesn't seem to work properly in this context... especially as a cron job } } else { if ($netdebug) { $sysstatus = system( "/var/www/CGI-Executables/dotcpdump stop echo" ) ;} if ( $mylast =~ "connect error" ) { exit 0 ; # prevents the stream of emails } else { open( LOGFH, '>>'.$logpath.'ln-mod-error' ) or die "Can't open mod error file: $! while reporting connect error \n"; # print "Failed to connect to I/O module at ".$timestamp."\n"; print LOGFH "Can't connect to I/O module after multiple attempts. $timestamp\n"; close LOGFH; open( LOGFH, '>>'.$lnhistory ) or die "Can't open history log file: $! while reporting connect error"; print LOGFH "Can't connect to I/O module after multiple attempts. $timestamp\n"; close LOGFH; # When the network comes back up, an idle state entr: should be appended to the log file. open( XFH, '>'.$snapshot ) or die "Can't open snapshot file $snapshot for write: $!\n"; print XFH "connect error ".$timestamp.$version ; close XFH; die "Can't connect to I/O module after $trials attempts. \n" ; } } select( undef, undef, undef, $trials-$trial ); } # print "Connected.\n"; # Should have generated a 3 packet [SYN] transaction $timestamp = localtime(); # a time stamp very close to the actual read time. my $trn = $modbus->transaction( $trs, $req ); # $req asks for 12 bytes in the read-coils request. 16 bytes are returned. my $res = $trn->execute(); my @mypat = @{ $res->inputs() }; my $mybits = join( "", @mypat[0..11] ); # Trim the vector to the fields actually returned. if ( 16 != @mypat ) { if ( $netdebug ) { system( "dotcpdump stop echo" ); } die "Received @mypat ". @mypat ." bytes. 16 bytes expected. \n"; } # print "data bytes: [@mypat] Reading >".$mybits."< Last reading %$mylast% at $timestamp \n"; open( XFH, '>'.$snapshot ) or die "Can't open snapshot file $snapshot for write: $!"; print XFH "$mybits $timestamp$version" ; close XFH; if ( not $mylast =~ $mybits ) { # Integrity check. $res = $trn->execute(); my $rebits = join( "", @{ $res->inputs() }[0..11] ); # print "reading $mybits; check rearad $rebits.\n"; if ( not $mybits =~ $rebits ){ open( LOGFH, '>>'.$logpath.'ln-mod-error' ) or die "Can't open mod error file: $!"; print LOGFH "pattern $mybits check-pattern $rebits $timestamp\n"; close LOGFH; carp ( "Alert: reading pattern $mybits check-pattern $rebits $timestamp\n" ); } } # print ' mylast '. $mylast. ' mybits '.$mybits.' file '.$lnhistory." \n"; if ( not $mylast =~ $mybits ) { # The place to recognize particular state transitions... like fill ended... my $flags = ( $mybits =~ ".{9}0.{2}" ) ? 'L':''; # Recognize large dewar fill $flags .= ( $mybits =~ ".{8}00.{2}" ) ? '*':''; # Recognize tape on the start switch! '01' and '10' are normal $flags .= ( $mybits =~ ".{10}0.{1}" ) ? 'S':''; # Recognize small dewar fill open( LOGFH, '>>'.$lnhistory ) or die "Can't open history log file: $!"; print LOGFH $mybits." ".$timestamp.$flags."\n"; close LOGFH; } open MYFH, ">".$lndata or die "couldn't open $!"; my $index=0; foreach ( @mypat[0..11] ) { # Build record for the CGI script to spit out print MYFH $features[$index],'§',(split(/#/,$featureStates[$index]))[$_],"#"; $index++; } print MYFH '['.$timestamp.']'."#"; close MYFH; print "$mybits at $timestamp\n";
Download test43-1430.jpg
image/jpeg 188.4k
test43-1430.jpg
Subject: Re: [rt.cpan.org #51251] /Library/Perl/5.8.8/Protocol/Modbus/Transport/TCP.pm uninitialized value error
Date: Mon, 09 Nov 2009 16:14:34 -0800
To: bug-Protocol-Modbus [...] rt.cpan.org
From: gtp <gtp [...] lbl.gov>
Download (untitled) / with headers
text/plain 513b
I found a place where in a match where a "\s+" had become a "\+" It was in the test after the ping command... so the script would not have detected an 'unreachable' host. The script then tried to connect to that unreachable host which resulted in my being sent me the line 27 error in .../TCP.pm. jerry Cosimo Streppone via RT wrote: Show quoted text
> <URL: https://rt.cpan.org/Ticket/Display.html?id=51251 > > > Can you please confirm that you are using the latest CPAN version > (0.05)? perl is 5.8.8. What platform? >
Download (untitled) / with headers
text/plain 276b
I'm glad you use Protocol::Modbus at Berkeley. If I ever come to the US, I'd like to see it :) Sorry, I'm a bit busy right now. As soon as I get some time, I shall change the TCP driver to check that the connection has been properly established. Thanks for the feedback.
Subject: Re: [rt.cpan.org #51251] - - comment
Date: Sat, 21 Nov 2009 13:05:58 -0800
To: bug-Protocol-Modbus [...] rt.cpan.org
From: gtp <gtp [...] lbl.gov>
Download (untitled) / with headers
text/plain 1.2k
Thanks for providing Protocol::Modbus. As a Modbus user, I'm a really small player with only 2 I/O modules. My project went with Perl because I knew some Perl and no JAVA when I started... and I thought I thought it might be easier to do the simple cgi web page stuff with it. (http://ln2.physics.berkeley.edu/status) When my network drops out I get errors when attempting to open a connections. That suggests a test condition... namely attempting to talk to a non-existent module. :) From here in Berkeley we have a wonderful view of the San Fransisco bay, San Fransisco, Alcatraz island, and Marin county. If you do come to this part of our state, consider putting the Muir Woods national park, in Marin County north of San Fransisco, on your itinerary. We think it's one of the local treasures. UC Berkeley has an art museum, and an exhibits in the anthropology building. Regards, Jerry Cosimo Streppone via RT wrote: Show quoted text
> <URL: https://rt.cpan.org/Ticket/Display.html?id=51251 > > > I'm glad you use Protocol::Modbus at Berkeley. If I ever come to the > US, I'd like to see it :) > > Sorry, I'm a bit busy right now. As soon as I get some time, I shall > change the TCP driver to check that the connection has been properly > established. > > Thanks for the feedback. >
Download (untitled) / with headers
text/plain 1.2k
Hi Gerald, I reviewed the code and from your logs I can see that the TCP connection failure is handled "correctly". That is, Protocol::Modbus is detecting that the TCP connection to your module failed, and it's reporting the failure through the connect() call down to the $transaction->execute() call. In my opinion, this is the correct behavior. So, what can we do to prevent this from happening? I would suggest slightly modifying your code: $timestamp = localtime(); # a time stamp very close to the actual read time. my $trn = $modbus->transaction( $trs, $req ); # $req asks for 12 bytes in the read-coils request. 16 bytes are returned. my $res = $trn->execute(); ... like this: $timestamp = localtime(); # a time stamp very close to the actual read time. my $trn = $modbus->transaction( $trs, $req ); # $req asks for 12 bytes in the read-coils request. 16 bytes are returned. my $res = $trn->execute(); # Connection failed? Try reconnecting and reissuing transaction my $retries = 3; while (! $res && $retries--) { sleep 1; $res = $trn->execute(); } ... So, you can retry the connection and the transaction issuing code until it succeeds. Let me know if this is useful. If it proves to be useful, maybe it's a good idea to integrate it in the execute() method itself.
RT-Send-CC: gtp [...] lbl.gov
Download (untitled) / with headers
text/plain 1.4k
On Sun Dec 20 09:46:51 2009, COSIMO wrote: Show quoted text
> Hi Gerald, > > I reviewed the code and from your logs I can see that the TCP > connection failure is handled "correctly". > > That is, Protocol::Modbus is detecting that the TCP connection to
your Show quoted text
> module failed, and it's reporting the failure through the connect() > call down to the $transaction->execute() call. > > In my opinion, this is the correct behavior. So, what can we do to > prevent this from happening? I would suggest slightly modifying your > code: > > $timestamp = localtime(); # a time stamp very close to the actual > read time. > my $trn = $modbus->transaction( $trs, $req ); # $req asks for 12 > bytes in the read-coils request. 16 bytes are returned. > my $res = $trn->execute(); > ... > > like this: > > $timestamp = localtime(); # a time stamp very close to the actual > read time. > my $trn = $modbus->transaction( $trs, $req ); # $req asks for 12 > bytes in the read-coils request. 16 bytes are returned. > my $res = $trn->execute(); > > # Connection failed? Try reconnecting and reissuing transaction > my $retries = 3; > while (! $res && $retries--) { > sleep 1; > $res = $trn->execute(); > } > > ... > > So, you can retry the connection and the transaction issuing code
until Show quoted text
> it succeeds. Let me know if this is useful. > If it proves to be useful, maybe it's a good idea to integrate it in > the execute() method itself.
Ping? Did you have a chance to try this?


This service is sponsored and maintained by Best Practical Solutions and runs on Perl.org infrastructure.

Please report any issues with rt.cpan.org to rt-cpan-admin@bestpractical.com.