This web page contains FAQ's for end users
FAQ's for Service Providers can be found here

  1.      Why the name CASTOR?

    The CASTOR Project stands for CERN Advanced STORage Manager, and its goal is to handle LHC data in a fully distributed environment.

  2.     From where can I download CASTOR?

    In this link you have the relevant links to download CASTOR (different releases) and create and upgrade DataBases.

  3.     Which OS and platforms are supported?

    Linux, distribution SLC4 and SLC5, 32 bits and 64 bits.

  4.     How to use CASTOR2 client

    Once the client is properly installed the next step is to set some enviroment variable to specify which stager your client will use and that you want to run CASTOR version 2.

      • STAGE_HOST should contain a valid name of a stager, you can get them here
      • RFIO_USE_CASTOR_V2 should be set to YES (soon to be dropped).

  5.     How to setup a CASTOR2 instance

    All spets detailled in this TWiki page

  6.     Stage not available

    If you get this error message: STG00 - stage daemon not available on castorlhcb (Connection refused)STG161 - Stage not available or in pause mode - Please wait Check that the value of RFIO_USE_CASTOR_V2 is YES (be sure that it is written with capital letters).

  7.     Unable to connect socket

    If you get this message Unable to connect socket to IP:PORT whenever you are using any stager commands, please check the value of the variable STAGE_HOST. This variable should exist and be initialized with an appropiate value.

  8.     Client and Server compatibilities.

    All CERN servers are running in version 2.1.1+

    Please notice that clients v2.0.* with servers v2.1+ require and specific version of castor-lib-compat (download page Savannah )

    CASTOR Client Version
    2.0.* 2.1.0.* >=2.1.1.*
    Server Version 2.0.* OK NO NO
    2.1.0.* OK with Castor-lib-compat-2.0.2-1 OK NO
    >=2.1.1.* OK with Castor-lib-compat-2.0.2-11 OK OK
  9.     Are there any limits on file size on CASTOR2 ?

    The castor2 (as well as castor1) software supports filesizes up to 2^64 bytes. However, the upper physical limit is set by the hardware resources used (consider the filesystems and the tapes sizes if you have your own castor instance to know the limit).If you are using CERN instances the recommended size is ~2-10GB. There is no lower limit imposed but we don't recommend to go below 0.1GB since the tape performance become very bad.

  10.     Firewall problems

    If rfcp does not work when your firewall is on but works when you disable it, you just have to open the range of ports [30000-30100] in your firewall. Alternatively, you can change/reduce/extend this range by modifying/adding these 2 lines in /etc/castor/castor.conf :

    • CLIENT LOWPORT 30000 # The port range to be used by client for
    • CLIENT HIGHPORT 30100 # the stager callbacks. If not present
  11.     Problems with rfcp

    If your having problems with rfcp command please follow the steps specified below:

    • Check the version of RFIO you are using and if it's located on /usr/bin/rfcp . You should used the RFIO provided by castor, if you use the one for dpm you won't be able to talk to castor for security policies.
    • Check that the right rmp is installed (castor-rfio-client-X.Y.Z-S).
      rpm -qa| grep rfio
  12.     Device or Resource Busy

    While writing a file the operation can be canceled due to a deliberate user action or an internal failure. Next time you will try to write again the file CASTOR won't allow you as CASTOR is wating for a put_done to close the file that was opened to write. In that situation thee file can be read but not duplicated, so when you try to read it from a different server as the file can not be duplicated you will receive "Device or Resource Busy".That error also appears if you try to write it again. Then in order to solve it check with "stager_qry" the status of the file and then execute manually the put_done, or remove it and write it again.

  13.     Client in a different network than stager and diskserver?

    If while executing a CASTOR command you received this messager "Host is not on local network and no mapping found" you have to check in which network the client is and if it's different to the network of the stager and the diskserver then add it to /etc/castor/castor.localhosts in the stager and in the diskserver.


    • Client IP
    • Stager IP
    • (In the stager) vi /etc/castor/castor.localhosts (add 137.138 or the given IP machine)

  14.     How to check tape status?

    Execute this sequence of commands:(But please remember that "showqueues should not be used for polling" )

        nsls -T file_aaa
        showqueues -x |grep VID_xxx
        vmgrlisttape -V VID_xxx

    More information can be found here

  15.    Error message Invalid argument

    If you are coping a remote file into a local file with rfcp command and you get the message "Invalid argument" Please check with nsls that the file exists and then check the permissions on the local directory where you are trying to write the file. The directory must be writable by your user.

  16.    stager_qry returns INVALID?

    If while using stager_qry you get the result "INVALID" then there is a problem with the location of the file due to :

    • file on a bad tape
    • disable tapes and inaccessible CASTOR file

    For more information check this twiki page

  17.     How can I debug an RFIO or a CASTOR client session?

    If you're experiencing troubles with your RFIO transfers, a way to get a more verbose trace of the process is by setting the following environment variables before issuing an rfcp command:


    RFIO_TRACE enables the trace of the RFIO transfer, whereas STAGER_TRACE enables the trace of the initial handshake with the CASTOR stager.

  18.     Under what license is CASTOR available?

    CASTOR is freely available under the terms of the OpenSource EU DataGrid software license.

  19.     How can I learn more on how to install and use CASTOR?

    You can follow one of the documentation/tutorials.


This document was last modified on:
For comments and changes on this page send email to the Webmaster
Copyright CERN