Description : AFD FAQ Author : Holger Kiehl email : Holger.Kiehl@dwd.de Version : 0.2 Last Updated: 06 August 2005 Copyright : GPL ====================================================================== Table of Contents 1. General Information 1.1 What is AFD? 1.2 Where can I get AFD? 1.3 Will AFD run with Windows XXX? 1.4 What is the history of AFD. 2. Compiling 2.1 How do I compile AFD under XXX? 2.2 Security considerations 2.2.1 Single user 2.2.2 Group access 2.2.3 Any users 2.3 I get the following error when compiling: 2.3.1 Can't open include file 'X11/Xaw/StripChart.h' 2.3.2 Can't find library for -lXpm (only versions < 1.2.2) 2.3.3 /usr/bin/ld: cannot find -lXm 3. Installing 3.1 How do I install AFD? 3.1.1 Installing from source 3.1.2 RPM 4. Configuration 4.1 When starting AFD or one of its process I get the following error: 4.1.1 ERROR : Failed to determine AFD working directory! 4.1.2 You are not permitted to use this program. [] 4.1.3 ERROR : Cannot read DIR_CONFIG file : No such file or directory 5. General usage 5.1 DIR_CONFIG/HOST_CONFIG 5.1.1 How do I activate the changes in the DIR_CONFIG or HOST_CONFIG file? 5.1.2 The password has changed but it is still seems to use the old one. What's wrong? 5.1.3 Do I always have to press the 'Reread HOST_CONFIG' and 'Reread DIR_CONFIG' button? 5.1.4 I have lost some or all the content of the DIR_CONFIG, what can I do? 5.2 Operational questions 5.2.1 The network connection to ALL hosts is down. What should be done? 5.2.2 The connection to one host is down and the job queue counter is constantly increasing. What action is necessary? 5.2.3 When a connection to one host is down should I stop transfer or stop the queue? 5.2.4 An error has occurred for connection XYZ. How do you know what type of error has occurred? 5.2.5 The problem to a host has been solved, but it always takes such a long time before it tries to send the files again. How can one test immediately whether the problem is solved or not? 5.2.6 There is no response or action at all. Even starting or stopping the AMG or FD does nothing. What's wrong? 5.2.7 There are files in the directory, but AFD does not send them. 6. Error Messages 6.1 FTP Error Messages 6.1.1 Disc quota exceeded. 6.1.2 Disc full. 6.1.3 The process cannot access the file because it is being used by another process. 6.1.4 Failed to send password for user XXXXX. 6.1.5 Failed to send password for user XXXXX due to timeout. 6.1.6 Failed to send password for user XXXXX. 530-User XXXXX: can't change directory to YYYY 6.1.7 Failed to connect() to XXXXX : Connection timed out. 6.1.8 Failed to connect() to XXXXX : Connection refused. 6.1.9 421-Service not available, closing control connection. 6.1.10 Remote hang up. ====================================================================== 1. General Information 1.1 What is AFD? The Automatic File Distributor provides a framework for very flexible, non-stop, log and debug-able delivery of an arbitrary amount of files to multiple recipients as expressed in URLs (currently mailing and ftp supported with the mailto://user@domain and ftp://user:password@host URL conventions). AFD stands for Automatic File Distributor. 1.2 Where can I get AFD? You can get information from the following locations: http://www.dwd.de/AFD - Homepage ftp://ftp.dwd.de/pub/afd - directory containing the stable releasese of AFD ftp://ftp.dwd.de/pub/afd/development - directory with unstable releases of AFD ftp://ftp.dwd.de/pub/afd/bin - here you will find precompiled binaries 1.3 Will AFD run with Windows XXX? No. Currently there is some work to get AFD running under Cygwin. 1.4 What is the history of AFD? In 1995 the German Weater Service (Deutscher Wetterdienst (DWD)) was doing its international file exchange via scripts. It was soon realized that this did cause a very high cpu load and was not very flexible. For this reason it was decided to develop a file distribution system that was to be very portable and easy to handle by a single person. Here follow some important millestones of the AFD: - 04.02.1996 First release and first operational use at DWD (only for distributing files to international partners). - 22.06.1998 Release of version 1.0.0 with a complete rewrite of the process FD to handle the internal communication between FD and AMG via FIFO's. - 08.08.1999 Released 1.1.0. Included time job and process to monitor other AFD's. - 03.02.2000 AFD released under the GENERAL PUBLIC LICENSE (GNU). - 24.09.2000 Released 1.2.0. AFD is now able to retrieve files via FTP from a remote host. ---------------------------------------------------------------------- 2. Compiling 2.1 How do I compile AFD under XXX? Run ./configure to generate config.h and the various Makefiles. ./configure --help gives a list of possible options with slightly longer descriptions in README.configure. See also 2.2 Security considerations. Some systems require unusual options for compilation or linking that the `configure' script does not know about. You can give `configure' initial values for variables by setting them in the environment. You can do that on the command line like this: ./configure CC=c89 CFLAGS=-O2 LIBS=-lposix Before compiling you might want set some other preferences in src/afdsetup.h and/or src/init_afd/afddefs.h. But note, changing some values here might make this incompatible with a previous AFD version and AFD will no longer work when it is started. So best to leave these values as they are. If you do change them it's best to run the following command before starting AFD again: afd -i This will initialize the internal database of AFD. After this it will be possible to run AFD with the new values. 2.2 Security considerations Before compiling AFD you need to decide where you want to install AFD and if you want other users to have access to your AFD. There are currently three possibilities: 2.2.1 Single user This is the default and you need to set no addititional flags with configure. Here only the user that has started the AFD may have access and may control it. 2.2.2 Group access Via this method all users of the same group have access and may control the AFD. This access can be further restricted in the $AFD_WORK_DIR/etc/afd.users file. To use this you need to start configure with the option --enable-group_can_write. 2.2.3 Any users To allow any user to have access you must set the flag --enable-setuid_progs in configure. This will set the setuid flag for some binaries. The access can be further restricted in the $AFD_WORK_DIR/etc/afd.users file. It is not recommendet to set this if you install the binaries as root. 2.3 I get the following error when compiling: 2.3.1 Can't open include file 'X11/Xaw/StripChart.h' HPUX for example does not include the Xaw libraries and headers. You can either install these on your system or in src/UI/Motif/Makefile.am comment out all occurance of afd_load. Then AFD will compile without the dialog afd_load, which is not required for running AFD. 2.3.2 Can't find library for -lXpm (only versions < 1.2.2) Most commercial unix vendors do not include the Xpm libraries. so either install these on your system or comment out the line: #define _WITH_XPM You can savely do this and you will hardly know any difference. 2.3.3 /usr/bin/ld: cannot find -lXm (only versions < 1.3.0) Either you do not have the Motif library or you need to supply the correct path to it in Include.mk with the variable XLIBPATH. On a Linux system this is for example set as follows: XLIBPATH = -L/usr/X11R6/lib If you do not have Motif install either OpenMotif or LessTiff. ---------------------------------------------------------------------- 3. Installing 3.1 How do I install AFD? 3.1.1 Installing from source In the source directory just issue the command 'make install' to install the binaries in the ../bin and ../sbin directories. 3.1.2 RPM This is the easies method, just issue the following command: rpm -Uvh To create an RPM, the simples way is to issue the following commands as root: rpm -Uvh cd /usr/src/redhat/SPECS # This is distribution dependent rpmbuild -ba afd.spec ---------------------------------------------------------------------- 4. Configuration 4.1 When starting AFD or one of its process I get the following error: 4.1.1 ERROR : Failed to determine AFD working directory! Either the environment variable AFD_WORK_DIR is not set or you did not supply the working directory of the AFD with the '-w' option. Without it AFD will not run. 4.1.2 You are not permitted to use this program. [] The user is not permitted to start the program. If you want to give the user the permission to start this program add the users name to $AFD_WORK_DIR/etc/afd.users and the programs he may use. 4.1.3 ERROR : Cannot read DIR_CONFIG file : No such file or directory There is no DIR_CONFIG file in the $AFD_WORK_DIR/etc directory. To run AFD you must have a valid DIR_CONFIG file! ---------------------------------------------------------------------- 5. General usage 5.1 DIR_CONFIG/HOST_CONFIG 5.1.1 How do I activate the changes in the DIR_CONFIG or HOST_CONFIG file? You need to press 'Reread DIR_CONFIG' or 'Reread HOST_CONFIG' in the afd_ctrl dialog under 'Control'. Or if you do not want to use the graphical interface use udc to update the DIR_CONFIG or uhc to update the HOST_CONFIG from the sbin directory. 5.1.2 The password has changed but it is still seems to use the old one. What's wrong? The message that is causing this error still has the old password. You must manually change this yourself (this will hopefully be fixed in Version 2.x.x). To get the message number look into the TRANSFER_LOG, you will see the number right after the '#' sign at the end of the error message. Now edit ths message with this number in the message directory and then press the 'Retry' button. 5.1.3 Do I always have to press the 'Reread HOST_CONFIG' and 'Reread DIR_CONFIG' button? No, if you have made changes to both the DIR_CONFIG and HOST_CONFIG file you just need to press the 'Reread DIR_CONFIG' button (or call udc from the sbin directory). 5.1.4 I have lost some or all the content of the DIR_CONFIG, what can I do? There is a tool get_dc_data which when called without arguments will print the whole DIR_CONFIG, as stored within AFD, to standard out. 5.2 Operational questions 5.2.1 The network connection to ALL hosts is down. What should be done? No action is necessary. AFD will stop the queue those hosts where a lot of jobs queue up and it will start them again as soon as the network comes back. If unsure you can stop AMG so no more jobs will be generated for FD. FD will still continue to try and distribute files at regular intervalls. 5.2.2 The connection to one host is down and the job queue counter is constantly increasing. What action is necessary? Actually no action is necessary. Since the AFD will itself stop the queue for this host if it reaches any critical value (and it will automatically reactivate the queue when everything is okay again). But it also does not hurt when you stop the queue for this host. But don't forget to turn it on again! 5.2.3 When a connection to one host is down should I stop transfer or stop the queue? It depends on how long the connection to host is down and on the number of files that are to be generated. Stop the transfer if you wish to interrupt the current transfers and/or you know this is only a short interrupt. But be careful that not to many jobs queue up (AFD will automatically stop the queue if this is the case and start it again). Best thing is to just stop the queue. Or leave it to AFD, it will automatically stop the queue and start it again. 5.2.4 An error has occurred for connection XYZ. How do you know what type of error has occurred? Look in the Transfer Log file and see what type of error it is. If it says "Failed to connect." then it is a network error or the remote host is down. Check the network to the remote host. With other errors it is best to turn on the debug mode for this host and evaluate the output in the Transfer Debug Log. (see also 6.1) 5.2.5 The problem to a host has been solved, but it always takes such a long time before it tries to send the files again. How can one test immediately whether the problem is solved or not? Press the retry button. 5.2.6 There is no response or action at all. Even starting or stopping the AMG or FD does nothing. What's wrong? Most properly the AFD was not shutdown correctly. Log out of the afd_ctrl and log in again. Be patience this can take a while (up to 30s) because every time you call the program afd it will check whether there is an AFD already running. If not it will restart the AFD. All other programs and shared memory areas will be removed. 5.2.7 There are files in the directory, but AFD does not send them. There could be many reasons why the files are not being send: - Make sure that there is a rule in the DIR_CONFIG to transmitt those files. - The files in the source directory must have the correct permission, ie. AFD must have read permission for those files. - Ensure that the input queue or transfer is NOT stopped in the afd_ctrl dialog. The two leds next to the hostname and debug led must be green. ---------------------------------------------------------------------- 6. Error Messages 6.1 FTP Error Messages 6.1.1 Disc quota exceeded. The user may not write any more data into this directory since since the quota is exceeded. 6.1.2 Disc full. The disk on the remote host is full. 6.1.3 The process cannot access the file because it is being used by another process. Another process on the remote machine has the file we wish to send open and the operating system does not allow the ftp daemon to open the same file again. 6.1.4 Failed to send password for user XXXXX. The password for user XXXXX is wrong. To correct this correct the password in the DIR_CONFIG and read 5.1.2. 6.1.5 Failed to send password for user XXXXX due to timeout. If this error message appears frequently then its very likely that the remote machine is overloaded and not able to handle your request. This is a common error on computers with NT. If this message appears only once then the network connection is broken. 6.1.6 Failed to send password for user XXXXX. 530-User XXXXX: can't change directory to YYYY The directory YYYY for user XXXXX on the remote host, does not exist or does not have the correct permissions. 6.1.7 Failed to connect() to XXXXX : Connection timed out. The remote host is not reachable. Either the network to this host is not functioning properly or the host is down. This seems to be a very common error. 6.1.8 Failed to connect() to XXXXX : Connection refused. Here we do reach the remote host via the network but there is no FTP daemon running or listining at the expected port. 6.1.9 421-Service not available, closing control connection. This error message can appear after we connected to the remote host. It means that it is currently being shuttown and not able to handle any further requests. 6.1.10 Remote hang up. If you only receive this error it means that the remote host refuses any connections to this port. It could be an indication that the remote host is overloaded or is in big trouble. If this error appears in conjunction with another error message, for example: Failed to change directory to YYYY. Then the remote host has just cut the connection, either because it is being shuttdown or has a very high load.