Whole document tree
    

Whole document tree

Debugging, Tracing, and Log Files

4.21. Debugging, Tracing, and Log Files

The LPRng software was designed and written to provide as high a level of diagnostic information as possible. This was largely in part due to the problems with portability, coding errors, and other human frailties. Approximately 80% of the LPRng source code concerns itself with checking return values from system functions and producing error messages, debugging and tracing information, and various facilities used for regression testing and diagnosis.

The approach used by LPRng is to produce trace output for the LPRng clients or log files for the lpd server that show the various events or flow of information through the LPRng system. There are several classes or types of actions that can be traced, and various levels of trace information generated. The interface used to control these actions are the command line -D literals flags and the lpc debug command.

First, we will look at how you can use the debugging facilities for the clients. Enter the following commands:

    h4: {292} % lpr -D=
    debug usage: -D [ num | key=num | key=str | flag | flag@ | flag+N ]*
      keys recognized: network[+N,@], database[+N,@], lpr[+N,@],
       lpc[+N,@], lprm[+N,@], lpq[+N,@], test=num, job=num, log[+N,@]
    h4: {293} % lpr -V /tmp/hi
    LPRng-3.7.2, Copyright 1988-2000 Patrick Powell, <papowell@lprng.com>
    sending job 'papowell@h4+981' to lp@localhost
    connecting to 'localhost', attempt 1
    connected to 'localhost'
    requesting printer lp@localhost
    sending control file 'cfA981h4.private' to lp@localhost
    completed sending 'cfA981h4.private' to lp@localhost
    sending data file 'dfA981h4.private' to lp@localhost
    completed sending 'dfA981h4.private' to lp@localhost
    done job 'papowell@h4+981' transfer to lp@localhost
    h4: {294} % lpr -D1 /tmp/hi
    09:38:08.707 h4 [13991] lpr  Get_printer: original printer '<NULL>'
    09:38:08.708 h4 [13991] lpr  Get_all_printcap_entries: starting
    09:38:08.708 h4 [13991] lpr  Select_pc_info: looking for 'all', depth 0
    09:38:08.708 h4 [13991] lpr  Select_pc_info: returning '<NULL>'
    09:38:08.708 h4 [13991] lpr  Select_pc_info: looking for '*', depth 0
    09:38:08.708 h4 [13991] lpr  Select_pc_info: returning '<NULL>'
    09:38:08.708 h4 [13991] lpr  Dump_line_list: Get_all_printcap_entries
    ...


The lpr -D= causes the lpr (and other LPRng programs) to show what debugging flags are available. The lpr -V flag causes lpr to run in verbose mode and show its activities. Finally, we use lpr -D1 to enable the simplest level of debugging. This will produce a trace of the various activities that lpr carries out. Try lpr -D2, lpr -D3, and so forth to see the increasing amount of detail that you get.

The network and database debug flags turn on debugging for the network facilities and the database (lpd.conf, printcap, and lpd.perms) lookups. Lets see what lpr -Dnetwork shows us:

    Ch4: {295} % lpr -Dnetwork /tmp/hi
    lp: getconnection: START host localhost, timeout 10, connection_type 1
    lp: getconnection: fqdn found localhost.private, h_addr_list count 1
    lp: Link_dest_port_num: port 2000 = 2000
    lp: getconnection: sock 3, src ip 127.0.0.1, port 65209
    lp: getconnection: dest ip 127.0.0.1, port 2000
    lp: getconnection: connection to 'localhost' socket 3, errormsg 'No Error'
    lp: Link_send: host 'localhost' socket 3, timeout 6000
    lp: Link_send: str '^Blp
    ', count 4, ack 0x80447a0
    lp: Link_send: final status NO ERROR
    lp: Link_send: host 'localhost' socket 3, timeout 6000
    lp: Link_send: str '^B135 cfA276h4.private
    ', count 22, ack 0x8044370
    lp: Link_send: final status NO ERROR
    lp: Link_send: host 'localhost' socket 3, timeout 6000
    lp: Link_send: str 'Hh4.private
    Ppapowell
    J/tmp/hi
    CA
    Lpapowell
    Apapowell@h4+276
    D2000-06-02-09:44:52.369
    Qlp
    N/tmp/hi
    fdfA276h4.private
    UdfA276h4.private
    ', count 136, ack 0x8044370
    lp: Link_send: final status NO ERROR
    lp: Link_send: host 'localhost' socket 3, timeout 6000
    lp: Link_send: str '^C3 dfA276h4.private
    ', count 20, ack 0x8044310
    lp: Link_send: final status NO ERROR
    lp: Link_send: host 'localhost' socket 3, timeout 6000
    lp: Link_send: str '', count 1, ack 0x8044310
    lp: Link_send: final status NO ERROR


As we see, we get a detailed exposition of the network connection and transfer steps. If you need or want more detail, try using lpr -Dnetwork+2 or lpr -Dnetwork+3. You may want to try lpr -Ddatabase and observe the actions of the lpr program as it extracts information from the lpd.conf and printcap files. If you need or want more detail, try using lpr -Ddatabase+2 or lpr -Ddatabase+3.

If you need to trace the activities of the lpd server, it becomes a little more complex. The lpd server has a single listening process that forks and creates individual processes to handle incoming requests. Debug or diagnose the main process actions by using lpd -D.... You may also want to use lpd -F to keep the server in the foreground so you can kill it off easily. Needless to say, you should also redirect the STDERR and STDOUT so that it goes to a file so that you can examine the voluminous records at your leisure. The following shows a typical main lpd process debugging session using the C Shell.

    h4: {296} % lpd -F -D1 >&/tmp/logfile &
    [2] 14299
    h4: {297} % tail -f /tmp/logfile
    2000-06-02-09:53:39.716 h4 [1200] Waiting Read_server_status: \
                             select status 1
    2000-06-02-09:53:39.716 h4 [1200] Waiting Read_server_status: \
                             read status 1
    2000-06-02-09:53:39.716 h4 [1200] Waiting Dump_line_list: \
        Read_server_status - input - 0x8047980, count 0, max 0, list 0x0
    2000-06-02-09:53:39.716 h4 [1200] Waiting Read_server_status: \
                             select status 0
    2000-06-02-09:53:39.716 h4 [1200] Waiting lpd: LOOP START
    2000-06-02-09:53:39.716 h4 [1200] Waiting Get_max_servers: \ 
                             getrlimit returns 64
    2000-06-02-09:53:39.716 h4 [1200] Waiting Get_max_servers: \
                             returning 32
    2000-06-02-09:53:39.716 h4 [1200] Waiting lpd: \
                             max_servers 32, active 0
    2000-06-02-09:53:39.716 h4 [1200] Waiting lpd: \
                             starting select timeout 'yes'
    
    ^C
    h4: {298} % jobs
    [1]  - Running                lpd -F -D1 >& /tmp/logfile
    h4: {299} % kill %1


We start the debugging session by running the lpd server in foreground mode. This causes it to send its output to STDOUT and STDERR. We redirect both of these to a file and put the lpd server in the background. Then we use tail -f to read from the log file. Finally, we kill off the lpd server.

This method is extremely difficult to use, as all of the output produced by the server and its subprocesses is sent to a single output file. If we want to debug the actions concerning a single queue, then we can use the queue log file and lpc debug command instead. The following options control debugging of an individual print queue.

lf=log

The log file for the queue. The queue server process will open this file and place debugging information into this file.

max_log_file_size=nnn

The maximum size of the log file in K bytes. When the queue server process first opens this file it will check to see if the file is larger than the maximum size. If it is, then it will truncate it. A zero (0) value suppress truncation.

min_log_file_size=nnn

When the log file is truncated only the the last nnn K bytes are retained.

db=options

These are debugging options for the spool queue. These options are permanent and cannot be changed by using the lpc debug facility.



The lpc debug command is used to set the debugging options in force for the spool queue. This is done by writing the debug options into the spool queue control file. Let us see how we can use this facility to trace the actions of printing a file.

Edit the printcap file so it have the contents indicated below, create the /tmp/lp and /tmp/lp2 files with 0777 permissions. Use checkpc -f to check the printcap, and then use lpc reread to restart the lpd server.

    # printcap
    lp:force_localhost
    lp:server
      :lp=/dev/null
      :sd=/var/spool/lpd/%P
      :lf=log
    lp2:force_localhost
    lp2:server
      :sd=/var/spool/lpd/%P
      :lp=/tmp/lp2
      :lf=log


Now execute the following commands:

    h4: {300} % lpq
    Printer: lp@h4
     Queue: no printable jobs in queue
    h4: {301} % lpc debug lp 1
    Printer: lp@h4
    debugging override set to '1'
    lp@h4.private: updated
    h4: {302} % lpc status
     Printer Printing Spooling Jobs Server Subserver Redirect Status/(Debug)
    lp@h4    enabled  enabled    0    none    none          (1)
    h4: {303} % lpr /tmp/hi
    h4: {304} % more /var/spool/lpd/lp2/log
    2000-06-02-10:10:50.589 h4 [1201] (Server) lp: \
       Update_spool_info: printer 'lp'
    2000-06-02-10:10:50.590 h4 [1201] (Server) lp: \
       Do_queue_jobs: printable 1, held 0, move 0
    2000-06-02-10:10:50.590 h4 [1201] (Server) lp: \
       Do_queue_jobs: after Scan_queue next fd 5
    2000-06-02-10:10:50.590 h4 [1201] (Server) lp: \
       Do_queue_jobs: MAIN LOOP
    2000-06-02-10:10:50.590 h4 [1201] (Server) lp: \
       Do_queue_jobs: Susr1 before scan 0
    2000-06-02-10:10:50.591 h4 [1201] (Server) lp: \
       Do_queue_jobs: chooser '<NULL>', chooser_routine 0


The lpc debug command sets the debug level to 1. We can use the lpc status command to see what debug flags or actions are currently specified for the spool queue. We then send a job to the spool queue and examine the log file contents.

Each line in the log file has a timestamp, the name of the host, the process id that produced it, and a heading that tells the action or activity that the process is performing, and the name of the print queue that is being processed and a trace message. By convention, the trace message lists the name of the routine that processed it and then the actual information. Some messages may extend over several lines, but each line has the standard header at the start of the line.

The default debug or trace actions were designed to trace problems with printing, as these are the most common. However, you can also use the lpr, lpc, lprm, or lpq option to cause the lpd server to trace the actions during the execution of an lpr, lpc, lprm, or lpq request.

The log option is used to test various logging facilities and is usually not used for general purpose debugging.