|
|
| SMS Server Tools 3 Community |
| Welcome, Guest. Please login or register. |
Sat Apr 27, 2024 12:05 |
This topic is locked
Page: 1
| Author |
Post |
|
|
#1 Thu Nov 19, 2009 09:03, 175 months ago.
|
|
Member
Registered:
Oct 2009
Location: Latvia
|
Operating system name and version: Debian Version of smsd: 3.1.5 Smsd installed from: sources Name and model of a modem / phone: Nokia 30 Interface: serial with USB adapter I have rear but regular problems ~ once a week when it looks like smsd does not work. The last record in log file was: SMS received (part 1/2) From: ... From other side htop shows smsd in process list.  My modem leds shows "Message received" (green-red blink of third led). After smsd restart all goes fine. It recieves second part of message and sends out unsent messages that was on phone memory ( I think so that I had no files in outgoing folder). Is it bug or I can use setting: receive_before_send = yes How to track out the problem? Increase loglevel? brgds Janeks
|
|
|
#2 Thu Nov 19, 2009 09:07, 175 months ago.
|
|
Member
Registered:
Oct 2009
Location: Latvia
Topic owner
|
Hm...
When I looked into htop after restart there was two smsd processes: mainprocess and GSM1
Before (when problems) was only one.
Janeks
|
|
|
#3 Thu Nov 19, 2009 10:11, 175 months ago.
|
|
Administrator
Registered:
May 2009
Location: Jyväskylä, Finland
|
Perhaps this has something to do with multipart messages, do you often receive them, more than once per week?
There should always be two processes listed. When this failure happens, you could use ps -efj list to check if a modem process is in a <defunct> state.
After a second part is received, the message is complete, or is it broken somehow?
Smsd does not send unsent messages from the phones memory. The phone will do it by itself.
I do not know if there is a this kind of bug. However it seems that something abnormal has happened.
You could set loglevel = 7 and collect the log until this failure happens again. Then show the tail of a log with all lines which are related to the last couple of minutes. Also show the content of smsd.conf file.
|
|
|
#4 Thu Nov 19, 2009 10:32, 175 months ago.
|
|
Member
Registered:
Oct 2009
Location: Latvia
Topic owner
|
I am getting multipart messages more offten if not all the time.
The message received upon failure is o'k.
I raised loglevel - let's see...
|
|
|
#5 Thu Nov 19, 2009 10:43, 175 months ago.
|
|
Administrator
Registered:
May 2009
Location: Jyväskylä, Finland
|
One more thing: you have used this same modem, adapter and version of smsd from at least 5 weeks ago. Can you locate a date when the problem started to happen? If the problem is newer than 5 weeks, probably you could show your current eventhandler code.
|
|
|
#6 Fri Nov 20, 2009 09:31, 175 months ago.
|
|
Member
Registered:
Oct 2009
Location: Latvia
Topic owner
|
Here is the eventhandler: #!/bin/bash
if [ "$1" == "RECEIVED" ]; then
# Conversion to unicode
if sed -e '/^$/ q' < "$2" | grep "^Alphabet: UCS2" > /dev/null; then
TMPFILE=`mktemp /tmp/smsd_XXXXXX`
sed -e '/^$/ q' < "$2" | sed -e 's/Alphabet: UCS2/Alphabet: UTF-8/g' > $TMPFILE
sed -e '1,/^$/ d' < "$2" | iconv -f UNICODEBIG -t UTF-8 >> $TMPFILE
mv $TMPFILE "$2"
fi
fi
if [ "$1" == "SENT" ]; then
# Conversion to unicode
if sed -e '/^$/ q' < "$2" | grep "^Alphabet: UCS2" > /dev/null; then
TMPFILE=`mktemp /tmp/smsd_XXXXXX`
sed -e '/^$/ q' < "$2" | sed -e 's/Alphabet: UCS2/Alphabet: UTF-8/g' > $TMPFILE
sed -e '1,/^$/ d' < "$2" | iconv -f UNICODEBIG -t UTF-8 >> $TMPFILE
mv $TMPFILE "$2"
fi
fi
smsrebproc "$1" "$2" "$3" > /home/storage/smsproc/testproc.txt
# This is an example script that logs all events into an SQL database
# You need a MYSQL database as described in the documentation.
# Please read the documentation before using this script.
SQL_HOST=localhost
SQL_USER=******
SQL_PASSWORD="*********"
SQL_DATABASE=******
SQL_TABLE=*******
DATE=`date +"%Y-%m-%d %H:%M:%S"`
#Extract data from the SMS file
FROM=`formail -zx From: < $2 | sed 's/"//g'`
TO=`formail -zx To: < $2`
#Remove plus sign, spaces, minus and short number prefix
TO=`echo "$TO" | sed 's/ //g' | sed 's/+//g' | sed 's/s//g' | sed 's/-//g'`
SUBJECT=`formail -zx Subject: < $2`
SENT=`formail -zx Sent: < $2`
TEXT=`sed -e '1,/^$/ d' < $2`
TEXTSQL=$(echo "$TEXT" | sed 's/\\/\\\\/g' | sed 's/\"/\\\"/g')
#Set some SQL parameters
if [ "$SQL_PASSWORD" != "" ]; then
SQL_ARGS="-p$SQL_PASSWORD";
else
SQL_ARGS="";
fi
SQL_ARGS="-h $SQL_HOST -u $SQL_USER $SQL_ARGS -D $SQL_DATABASE -s -e"
#Insert a new entry into the SQL table
if [ "$1" = "FAILED" ] || [ "$1" = "SENT" ]; then
mysql $SQL_ARGS "insert into $SQL_TABLE (type,sent_time,sender,receiver,msgid,text) values (\"$1\",\"$DATE\",\"$FROM\",\"$TO\",\"$3\",\"$TEXTSQL\");";
elif [ "$1" = "RECEIVED" ]; then
mysql $SQL_ARGS "insert into $SQL_TABLE (type,sent_time,received_time,sender,receiver,text) values (\"RECEIVED\",\"$SENT\",\"$DATE\",\"$FROM\",\"$SUBJECT\",\"$TEXTSQL\");";
elif [ "$1" = "REPORT" ]; then
#Extract more data from the status report file
DISCHARGE=`sed -e 1,/SMS\ STATUS/d < $2 | formail -zx Discharge_timestamp:`
MSGID=`sed -e 1,/SMS\ STATUS/d < $2 | formail -zx Message_id:`
STATUS=`sed -e 1,/SMS\ STATUS/d < $2 | formail -zx Status: | cut -f1 -d,`
if [ "$MSGID" != "" ]; then
ID=`mysql $SQL_ARGS "select id from $SQL_TABLE where receiver=\"$FROM\" and type=\"SENT\" and msgid=\"$MSGID\" order by id desc limit 1;"`
mysql $SQL_ARGS "update $SQL_TABLE set received_time=\"$DISCHARGE\",status=\"$STATUS\" where id=\"$ID\";"
fi
fi
'bash' Syntax Highlight powered by GeSHi |
|
|
#7 Fri Nov 20, 2009 10:22, 175 months ago.
|
|
Administrator
Registered:
May 2009
Location: Jyväskylä, Finland
|
Thank's. I was wondering if in some cases eventhandler never returns and causes hanging. But probably this is not the reason.
You are using internal_combine = yes setting. After the first part of a message was received, this part was handled but then smsd seems doing nothing. Because we do not have a detailed log of a hanging, it's difficult to know what smsd was last doing.
When you are now running with loglevel 7, we just have to wait until the problem occurs again. When that happens, a log file is very important but you could also show the content of a file /var/spool/sms/incoming/GSM1-concatenated. Probably the content is correct, but it's good to check anyway.
Can you locate some last problematic incoming message files? As the message was correct after restarting smsd and receiving the second part, the first part must have something which causes this problem. There is no PDU's stored in the message files, unless you have used store_received_pdu = 3 in the smsd.conf. If there are PDU's, they can be used for debugging. Depending of a message and it's sender, is it possible to re-generate that kind of a message and then get it received to see if the problem is repeated?
|
|
|
#8 Sun Nov 22, 2009 19:13, 175 months ago.
|
|
Member
Registered:
Oct 2009
Location: Latvia
Topic owner
|
Ups, I forgot to set those PDU files, but I have log records with loglevel = 7: You can note the problem place by time break. That's the time when I restarted smsd. « Last edit by keke on Sun Nov 22, 2009 19:23, 175 months ago. » |
|
|
#9 Sun Nov 22, 2009 19:31, 175 months ago.
|
|
Administrator
Registered:
May 2009
Location: Jyväskylä, Finland
|
Okay, it seems that there is a bug in the code  . I edited your log and will examine this tomorrow morning, currently I'm unable to check this more. While waiting for the solution, set the global setting in the smsd.conf ic_purge_hours = 0. |
|
|
#10 Mon Nov 23, 2009 12:45, 175 months ago.
|
|
Administrator
Registered:
May 2009
Location: Jyväskylä, Finland
|
I have tested this issue with your PDU's, but there was no troubles.
Can you show what you had in your /var/spool/sms/incoming/GSM1-concatenated file? Also, is there temporary file still remaining, ls -l /var/spool/sms/incoming/GSM1-concatenated* ? If there is, what is the content?
|
|
|
#11 Mon Nov 23, 2009 16:05, 175 months ago.
|
|
Member
Registered:
Oct 2009
Location: Latvia
Topic owner
|
I found that there is two files: One empty created at 2009-11-22 08:12:42 and another created at 21:04:53: GSM1.2009-11-22.Yvh7SL: |
|
|
#12 Mon Nov 30, 2009 09:36, 175 months ago.
|
|
Administrator
Registered:
May 2009
Location: Jyväskylä, Finland
|
Are you probably running 64 bit GNU/Linux and compiler?
|
|
|
#13 Mon Nov 30, 2009 10:29, 175 months ago.
|
|
Member
Registered:
Oct 2009
Location: Latvia
Topic owner
|
No, I am not!
The system where SMStools are running (and was compiled) is on PPC system.
But my laptop have 64bit.
It seems a bit strange that I did not get any hangups after changed that settings.
|
|
|
#14 Mon Nov 30, 2009 10:36, 175 months ago.
|
|
Administrator
Registered:
May 2009
Location: Jyväskylä, Finland
|
The setting ic_purge_hours = 0 disabled this feature.
I have to check this on PPC too when it's possible. If I cannot find any reason for this hanging, can I provide a test code for you to run it on your system?
|
|
|
#15 Mon Nov 30, 2009 10:51, 175 months ago.
|
|
Member
Registered:
Oct 2009
Location: Latvia
Topic owner
|
You could try.
If it is not very time consuming, than I can try to find time to do that.
|
|
|
#16 Mon Nov 30, 2009 18:42, 175 months ago.
|
|
Administrator
Registered:
May 2009
Location: Jyväskylä, Finland
|
Yeah, PPC was the key for this issue. With gcc version 4.3.2 (Debian 4.3.2-1.1) on PPC, the bug was clear and is now fixed. There is 3.1.6 available.
Thank you for the reporting.
|
|
|
#17 Mon Nov 30, 2009 19:30, 175 months ago.
|
|
Member
Registered:
Oct 2009
Location: Latvia
Topic owner
|
Thank you for your work!
|
This topic is locked
Page: 1
Time in this board is UTC.
|
|
|
 |
|
|
 |
|
