You need to be in the EY group to have write access to the disk. Ask Richard Dubois if you find you don't have write access.
sldtpwrite -v
TAPENAME -q1-
NFILE -lsl -FF -L32768 -gCART
FN1 FN2 ...
where:
TAPENAME = Silo tape name, eg qa0072
NFILE = # files to copy, eg 4
CART = a fixed keyword, not to be substituted
FN1 = filename of first disk file
FN2 = filename of second disk file
etc
Filenames need to be fully qualified, eg
/nfs/ravel/u1/richard/
MyNtuple.rzdat.
sldtpwrite is available from the SLD group area /afs/slac/g/sld/bin. You will need to have set up the standard SLD .sldcshrc environment to use sldtpwrite. It checks the SLD Oracle database to check ownership of the tape.
This command takes a bit of time, so it might be a good idea to run it in batch or in background. Be sure to keep an error file, and check that it runs OK.
stagein -V
TAPENAME -q
FILENUM
-FF -L32768 -s
MMM FN
where:
TAPENAME = Silo tapename, eg qq0072
FILENUM = file number on tape, eg 1
MMM = maximum disk file size in MB, eg 300
FN = filename which will be used for this data
Data will be copied from tape to the staging disk, which is dedicated to this useage. For normal operation, you do not need to know the actual name of this disk. There must be sufficient room on the staging disk for a file up to MMM MB in size; otherwise, stagein will fail. Hence, do not specify MMM much larger than it needs to be. I am not sure what happens if the actual file is longer than MMM MB.
The filename FN may point to any disk or partition,
even one with practically no free space (eg your home directory), because
stagein
does not actually put the data there. It will
simply contruct a
symbolic link from FN to wherever the file really is
on the staging disk.
When stagein returns, you can direct the analysis program
to use FN as input. Note that you can only access the
staged files from one of the interactive farm machines (currently the
Morgan0n machines).
This command may take some time to execute. If the file is already on disk (eg from an earlier stagein command by you or somebody else), stagein should know about it and immediately return with FN set up to link to the file.
The disk file will be purged by the staging system when it needs room for
other datasets; however, the link FN is your responsibility.
Use the rm
command to get rid of it. Note that the real disk file could be gone even
if you have a link pointing to it, i.e. always execute the stagein
command, and do not rely on the presence of a link.
stageclr -V
TAPENAME-q
FILENUM stageclr
your own files. Please do not
use the standard Unix command rm
for this purpose. The
space freed up may not be known to the staging system.
Use stageqry -s
to see how much space is available in the
staging system.
If you want to see which tapes are active, issue the command
tpstat
. Unlike the other commands, you must be logged onto
the Unix machine
known as silosrv1
.
stageout -V
TAPENAME -q
FILENUM -lnl -FF -L32768 -sMMM -poutput
FN ...
where:
TAPENAME = (arbitrary) tapename, e.g. FOO
FILENUM = (arbitrary) file number, eg 1
MMM = space to be reserved in MB, eg 300
FN = filename associated with it, e.g. BAR
When this command returns successfully, it will have created a link from FN to an actual disk location. You can then simply write to FN.
Note that you have reserved MMM MB of space, but your actual useage may be more or less. If you try to write more than the reserved space and if there is indeed more space available, it will work.