Accounts and Contact Information
|
| Q: |
Who should I contact if I want to use the computing resources at UWM? |
| A: |
Please fill out this form.
|
| Q: |
Can I post to the ldas-uwm or medusa-users mailing lists? |
| A: |
No, these lists are for announcements only. Questions about resources
that are not answered here should be sent to the medusa-admin mailing list at medusa-admin@gravity.phys.uwm.edu.
|
Getting Help
|
| Q: |
Who should I contact if I already have an account and I have a question about using the computing resources at UWM? |
| A: |
You should email the medusa-admin mailing list at
medusa-admin@gravity.phys.uwm.edu
|
| Q: |
Who should I contact if I am having problems debugging my code at UWM? |
| A: |
If you think that the issue is a result of the system configuration at UWM,
then you should mail the medusa-admin mailing list at medusa-admin@gravity.phys.uwm.edu. However if you are having problems with the development of your code, you should email the lal-discuss mailing list at lal-discuss@gravity.phys.uwm.edu.
|
| Q: |
How do I print out a documentation page? The text is too wide. |
| A: |
At the bottom of the page is a link saying "strip it for printing." Click
this link and you will get a version of the page without the extra clutter.
If it's still too wide to print, submit a bug report asking the maintainer
to trim the text.
|
Use of Resources
|
| Q: |
What machines can I log onto at UWM? |
| A: |
Unless you have express permission from your contact at UWM, you should
only use hydra.phys.uwm.edu to log into. Please respect this policy. If you
find this is holding up your project, then let us know and we will work with
you to implement a solution.
|
LIGO Frame Data
|
| Q: |
Is all the E7 data (supposed to be) on the UWM system now? |
| A: |
In order to make room for S1 data we removed E7 data from the LHO
interferometers. The E7 data from LLO is still available.
|
| Q: |
In analysing the E7 data from Livingston, I found a hole in the data from 693909088 to 693944176. Is this data missing? |
| A: |
Yes. The data does not exist anywhere.
From: Greg Mendell
To: duncan brown
Subject: Re: [LAL-Discuss] Fwd: Re: strange hole in the frames
I just checked that 693909088 to 693944176 is the missing 10 hours due
to the second framebuilder at LLO not writing after the Y2002 bug.
Cheers,
Greg
The missing data is Jan 01, 2002 08:31:15 UTC to Jan 01, 2002 18:16:03 UTC.
|
| Q: |
How do I build simulated frame files which can be analysed by LDAS at UWM? |
| A: |
This answer is not yet complete. I need to talk to Scott
and Patrick
You should use ascii2frame from Andre Vicere's
Fbe
package. This is also avaliable as part of
ligotools.
Generate your data as a single column plain text files containing the values
for the channels that you want to simulate (typically H1:LSC-AS_Q or
similar) and then use ascii2frame to create simulated frame
files.
You should adhere to the following guildlines for generating frame files at
UWM. First you should request a type code for
- The frame files should contain one frame per file of length 16 seconds.
This is what the frame builders at the sites produce, so we would like to stay
with this format.
- The frame files should be named correctly according to the official
naming convention, which is described below.
- You will
|
Using LDAS at UWM
|
| Q: |
What are the general rules for running LDAS jobs at UWM? |
| A: |
LDAS at UWM is a minimal installation designed for prototyping code.
Please do not dump MANY little jobs or any number of huge jobs on the system
at one time. Use Peter Shawhan's LDASjob
package to build pipelines which submit jobs sequentially if you want to
experiment on large amounts of data.
|
| Q: |
Can I submit a job from any computer or it do I have to submit my LDAS jobs from hydra? |
| A: |
You can submit LDAS user commands to the managerAPI on
ldas.phys.uwm.edu:10001 from any machine in the world. You do not
have to be on a UWM machine to submit an LDAS user command.
|
| Q: |
Would a datapipeline job find my shared objects under /home/username/ if the job was not submitted from UWM? |
| A: |
If you submit a datapipeline command and specify the -dynlib option
to point to a shared object in your home directory at UWM, the wrapperAPI will
find it, even if you are submitting the job remotely.
|
| Q: |
Does work guild with ldas-uwm? |
| A: |
Yes. ldas-uwm and all the databases at uwm are currently accessible via
guild. Just choose "UWM" from the Current LDAS server menu. We keep
Peter Shawhan up to date when we add and remove databases, so you should be
able to access all the existing databases at uwm via the Database
name menu.
|
| Q: |
Can I use the LDASjob package to submit jobs to UWM? |
| A: |
Yes. LDASjob
can be used from machines at UWM or from remote sites to submit jobs to the
managerAPI.
|
| Q: |
Where can I put response files for ldas-uwm to pick them up? |
| A: |
Unlike LDAS at the lab, all the LDAS APIs at UWM can see your home directory.
You can put response files anywhere under your home directory and access them
using the URL file:/home/myusername/path/to/responsefile.ilwd in your
LDAS user command.
|
| Q: |
Are there any release versions of lal or lalwrapper installed at UWM? |
| A: |
No, currently users must load their own shared objects from their home
directories by specifying the path in the -dynlib option of the
datapipeline command. This may change in the future as use of the system
evolves.
|
| Q: |
Can I use UNIX system calls such as printf() to help debug my code at UWM? |
| A: |
Yes. fprintf() statements and the like are OK at UWM, which is
currently a development and prototyping system, but NOT on the lab
systems. Make sure you remove these before committing your code to CVS.
Please be sparing with the number of debugging statements that you have, or it
may be difficult for yourself and other users to find the output in the
mpi.log file
|
| Q: |
Where can I access the stdout or stderr from my wrapperAPI job? |
| A: |
The stdout and stderr of all wrapperAPI jobs are dumped
to the file /ldas_outgoing/mpiAPI/mpi.log which should be readable by
all users with accounts on the system. Be aware that all the running
wrapperAPI jobs dump to that file simultaneously, so if several users are
writing to it, it may be difficult to find your debugging messages.
|
| Q: |
What databases can I use to store results from my jobs? |
| A: |
Use only the uwm_test database during prototyping. Do not
prototype with any other databases since they are being used for scientific
analyses at this time. Policies for database usage are being developed, but
until then we ask that you are circumspect about your use of anything except
uwm_test.
|
| Q: |
Is the data in the /ldas_outgoing/jobs/ directory backed up? |
| A: |
No. You should consider this to be volatile scratch space. Do not leave any
data there that you want to keep permanently.
|
| Q: |
What hardware do the UWM LDAS APIs run on? |
| A: |
Currently the APIs run on the following systems:
managerAPI
cntlmonAPI | Sun Ultra 30, 384Mb | Solaris 8 |
diskcacheAPI
frameAPI
dataconAPI | Intel P3 1Ghz, 1Gb RAM | Red Hat 7.1 |
mpiAPI
wrapperAPI | Intel P3 1Ghz, 512Mb RAM
medusa.phys.uwm.edu | Red Hat 6.2 |
eventmonAPI
ligolwAPI | AMD Athlon 1.2Ghz, 512Mb RAM | Red Hat 7.1 |
metadataAPI
IBM DB2 Database | Intel P3 1Ghz, 512Gb RAM | Red Hat 7.1 |
|
About this FAQ
|
| Q: |
What is the current CVS revision of FAQ? |
| A: |
This file has CVS id:
$Id: faq.txt,v 1.2 2003/05/23 19:18:11 duncan Exp $
|
| Q: |
Where is the program that generates this FAQ? |
| A: |
The HTML is generated by a CGI script that is located on gravity at
/var/www/cgi-bin/faqprocessor.cgi
|
| Q: |
How do I add questions to this FAQ? |
| A: |
Read the documentation instructions on the medusa web pages.
|