FAQs
In the following section, john
is used as an
example username, and 12345 is used as an example Job ID.
Be sure to replace these with the appropriate Michigan Tech
ISO username and Job ID respectively.
- When I try SSH into the cluster, I get something like
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @ WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED! @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ IT IS POSSIBLE THAT SOMEONE IS DOING SOMETHING NASTY! Someone could be eavesdropping on you right now (man-in-the-middle attack)! It is also possible that the RSA host key has just been changed. The fingerprint for the RSA key sent by the remote host is c6:90:98:f2:0b:a9:87:39:4c:a5:39:1e:0c:e9:6b:2b. Please contact your system administrator. Add correct host key in /home/john/.ssh/known_hosts to get rid of this message. Offending key in /home/john/.ssh/known_hosts:2 RSA host key has changed and you have requested strict checking. Host key verification failed.
If the SSH public key on your local machine was re-created OR if Information Technology Services specifically informed you to expect such a message -- because Rama was re-built and/or your account on Rama was re-created, then
- Open
/home/john/.ssh/known_hosts
- Search for rama.
- Delete the matching line(s).
- Save and close the file.
If that wasn’t the case, then contact Information Technology Services immediately [call (906) 487-1111 or emailit-help@mtu.edu
] with a clear description of this problem. - Open
- When I try SSH into the cluster, I get something like
Could not chdir to home directory /home/john: No such file or directory mkdir: cannot create directory `/home/john': No such file or directory It doesn't appear that you have set up your ssh key. This process will make the files: /home/john/.ssh/id_rsa.pub /home/john/.ssh/id_rsa /home/john/.ssh/authorized_keys Generating public/private rsa key pair. Enter file in which to save the key (/home/john/.ssh/id_rsa): Could not create directory '/home/john/.ssh'. Enter passphrase (empty for no passphrase): Enter same passphrase again: open /home/john/.ssh/id_rsa failed: No such file or directory. Saving the key failed: /home/john/.ssh/id_rsa. -bash: /home/john/.ssh/authorized_keys: No such file or directory chmod: cannot access `/home/john/.ssh/authorized_keys': No such file or directory chmod: cannot access `/home/john': No such file or directory -bash-3.2$
Send an email toit-help@mtu.edu
with a clear description of this problem. - I do not remember my password.
All users of Rama that belong to Michigan Tech may use their ISO password. If you do not remember your Michigan Tech ISO password, follow normal procedures for having it reset - by visiting either the IT User Services Help Desk in JRVP Library. If you do not belong to Michigan Tech and/or are away from the university, send an email toit-help@mtu.edu
with a clear description and your password will be reset. - I cannot SSH into the cluster and/or check its website
from off-campus.
Follow the instructions given here and use VPN. - I tried the solution in above FAQ and I still cannot
SSH into the cluster. My account seems to be locked.
Can I use some other user's credentials to log in?
No. Sharing login credentials (username and password) with other and/or unauthorized users is a violation of Acceptable Use of Information Technologies. Using this infrastructure in a manner that violates the above and/or other provisions described in the aforementioned policy will lead to immediate suspension of account to protect the integrity of the system and curtail abuse. To the extent computer usage is believed to be a violation of federal, state, or local laws, Michigan Tech will turn the matter over to the appropriate authorities. - I cannot run any/some program(s)!
You edited the contents above
# PLEASE DO NOT EDIT THE ABOVE CONTENTS
line in your login file or re-defined one of the variables already in use. If that wasn't the case, then send an email toit-help@mtu.edu
with a clear description of this problem (and name of the software). - How do I know which variables are already in use?
Run one of the the following commands
set
orenv
- Can I re-alias commands/variables to meet my needs?
Re-aliasing any of the variables (i.e., output ofset
orenv
commands) is bad idea. Doing so for any of theq*
commands counts as a violation of Rama's protocols and will be grounds for removal of your account. - When I submit a job via
qsub
command, I get
Unable to run job: denied: host "compute-M-N.local" is no submit host.
All jobs need to be submitted to the queue from the login node (also known as submission host) The error message appears when a job is submitted to the queue from a compute node (also known as an execution host), from within aqlogin
session. - Can I check my email (using
PINE
or something like that)?
No. Use your local machine for all activities beyond computing. - Will my data be backed up?
No. Users are responsible for their own data. - Can I run programs in the Front End/Login Node(s)?
No. Front End [in the absence of Login Node(s)] is intended for submitting jobs to the queuing system. Running simulations in Front End/Login Node(s) will slow the system down. All such simulations will be terminated without prior notice. Continued violation, after a warning, will lead to blocking/removal of account. - Where can I compile my code/software?
Use qlogin and it will log you into a compute/tile node to which you have access to. Standard set of compilers, tools and utilities will be available in that node, and you can compile your code. - When I run
screen
command in aqlogin
session, I get
Cannot open your terminal '/dev/pts/0' - please check.
Once in aqlogin
session, run
script /dev/null
and then run thescreen
- There is no job running on a particular compute node.
Can I log in and submit my job?
No. All jobs must ONLY be submitted via their respective SGE submission scripts from the Front End/Login Node(s) to the queuing system. - I submitted a job but can't find it now. There is (are)
no output file(s) either. What happened to my simulation?
Suppose that the script used to submit the job is calledprojectX.sh
. First, check if there is a file calledjob_info.12345
in the directory where you started your job. If so, it may contain some more information. Also, do not forget to checkprojectX.sh.o12345
file in the same location. If your job was aborted/suspended, you will have received an email from theroot
to that effect. - I submitted a job but it shows with
Eqw
status inquser
(orqstatus
). What should I do now?
First, learn more about the reasons forEqw
status by running the command
qstat -j 12345 -explain E
Try clearing the error (E
) state by running the command
qmod -cj 12345
If the job reverts back toEqw
status, then try deleting the job and re-submitting it to the queue. If the problem continues, then send an email toit-help@mtu.edu
with a clear description of this problem (and the relevant Job ID). - When I submit a job, the compute node asks for my password.
This probably means you are attempting run your calculation outside of the SGE queuing system. If that isn't the case, then send an email toit-help@mtu.edu
with a clear description of this problem (and name of the software). - Some commands do not work the way they should!
Read theman
page for that command. Meaning of a certain optional argument for the command might be different from the version/distribution of linux you use(d) elsewhere. - When trying to save files,
vi
says
E138: Can't write viminfo file /$HOME/.viminfo!
The following command should resolve this problemrm -f ~/.viminf?.tmp
If that did not solve the problem, then send an email toit-help@mtu.edu
with a clear description of this problem. - When I run
configure
script for some software, it picks up C/C++ and FORTRAN compilers from different makes (for e.g., GCC and Intel).
The output ofconfigure
command is to be treated as a best guestimate. When certain variables (for e.g.,CC
,CXX
,FC
,F77
,F90
, etc.) are set, it attempts picks up the right compilers, libraries and header files. While such attempts are mostly successful, often it is necessary to read through theMakefile
generated byconfigure
and make appropriate corrections before attempting themake
command. - Is peer-to-peer file sharing allowed?
No, if it is for personal and/or (non) commercial purposes. Yes, if and only if it is for academic and research purposes. However, distribution of such material is governed by the Digital Millennial Copyright Act. Michigan Tech's compliance with the Higher Education Opportunity Act and consequences of illegal file sharing are described here.
Information Technology Services will disable accounts of users found employing peer-to-peer file sharing for personal, non-academic, non-research and/or (non) commercial purposes in this computing infrastructure, and take additional action as necessary. - There seem to be a number
of unused CPU cycles. Can I use them for mining
BitCoins?
Yes, if and only if it is for academic and research purposes as indicated by your (or your advisor's) proposal/project - and has been granted explicit permission.
No, if it is for personal, non-academic, non-research and/or (non) commercial purposes. Information Technology Services will treat any such instance of BitCoin mining (and/or any derivative thereof and/or similar to) as a violation of Information Security Compliance Policy.
Any University employee, student or non-university individual with access to University data who engages in unauthorized use, disclosure, alteration, or destruction of data is in violation of this plan and will be subject to appropriate disciplinary action, including possible dismissal and/or legal action. - Can I run visualization programs -
for e.g., Jmol, MolDen, XCrySDen, etc.?
Yes and no. When explicitly permitted to do so, all such visualization programs must be run from within aqlogin
session. Length of such sessions and availability of programs is explained here. - The cluster's performance does not match my expectation
and/or that of some other cluster.
If you believe you have done everything you can to optimize your code (using the appropriate version of dependencies, relevant compilation flags, and variables in your SGE script) and still cannot get the performance you expect and/or you are used to getting in some other cluster, it is requested that you
- provide Information Technology Services with
information about
- name of and source code for your software as well as a complete list of its dependents (and their version).
- your methods of compilation - compiler and compilation flags used.
- your methods of running the software.
- metrics (specifically
tCPU
andtWall
) from Rama and that other cluster. - specifications (hardware, software, compilers, network, etc.) of that other cluster.
- put Information Technology Services in touch with administrators of that other cluster so that attempts can be to rebuild/customize the cluster in as much the same way as possible.
Information Technology Services is also more than willing to compile (and possibly optimize) your code. While every effort will be made to get this done as quickly as possible, owing to the very nature of compilation process, Information Technology Services cannot guarantee a set deadline. - provide Information Technology Services with
information about
- My problem does not seem to be listed. What should I do?
Read other FAQs to make sure that your problem is not a variant of some other problem. Results from a Google search often provide a quick solution. If these do not help, send an email toit-help@mtu.edu
with a clear description of the problem and Information Technology Services will work with you to diagnose/resolve the issue as soon as possible. - What do you mean by send an email to
it-help@mtu.edu
with a clear description of the problem?
It means that you should send an email to
it-help@mtu.edu
with a subject line that looks like
rama.research: issue title
and include as much information as possible; steps/commands that you have tried, as is error messages and screenshots will help diagnose/resolve the issue in a timely fashion.