Error installing 1.8.4

[laytonjbgmail]

Hello,

I tried installing 1.8.4 using the command,

REMORA_INSTALL_PREFIX=/home/laytonjb/bin/remora-1.8.4 ./install.sh

and I got an error:

ar rvs libsyscom.a common.o ioconf.o
ar: creating libsyscom.a
a - common.o
a - ioconf.o
gcc -o mpstat -g -O2 -Wall -Wstrict-prototypes -pipe -O2 mpstat.o librdstats_light.a libsyscom.a -s
Installing mpstat ...
./install.sh: 82: ./install.sh: Syntax error: Bad fd number
./install.sh: 77: [: unexpected operator
./install.sh: 88: ./install.sh: Syntax error: Bad fd number
./install.sh: 83: [: unexpected operator
./install.sh: 91: ./install.sh: Syntax error: "&" unexpected

BTW I'm using gcc -7.5.0 and gfortran 7.5.0 (this is Ubuntu 18.04).

In looking at the install.sh script is 1.8.4 targeted only at either Intel MPI or MVAPICH2?

Thanks!

Jeff

[milfeld]

Hi Jeff,
Thanks for reporting the problem.
Some external shell executions [e.g. x=$( ....)] were using the redirection " >& file" in install.sh.
This is offensive to some shells and results in the "Bad fd number".
In the install.sh script these were changed to >file 2>&1. This should fix the install problem.
FYI: if you don't have MPI or Lustre/IB you should truncate the modules list in src/config/modules
(after downloading and before executing install.sh) to have just this list:
cpu,CPU
memory,MEMORY
numa,NUMA
eth,NETWORK

You could also include
power,POWER
temperature,TEMPERATURE
but I'm not sure if these are generically running on other systems.

[laytonjbgmail]

I grabbed the new install.sh and no joy. The error looks similar as before:

ar rvs libsyscom.a common.o ioconf.o
ar: creating libsyscom.a
a - common.o
a - ioconf.o
gcc -o mpstat -g -O2 -Wall -Wstrict-prototypes -pipe -O2 mpstat.o librdstats_light.a libsyscom.a -s
Installing mpstat ...
./install.sh: 79: [: 0: unexpected operator
./install.sh: 85: [: 0: unexpected operator
./install.sh: 93: ./install.sh: Syntax error: "&" unexpected

I haven't tried changing the modules list - I'm going to try this next.

Thanks for the help!

Jeff

[milfeld]

Hi Jeff, Can you zoom, and share your screen on your machine? If you can, I've got a UT Zoom id so you can join a meeting. (Unfortunately, you have to create a zoom account to join, which is trivial, when it asks to create an account, just provide a name and password. It is that simple. ) Best, Kent https://utexas.zoom.us/j/98588550681

…

________________________________ From: laytonjbgmail <notifications@github.com> Sent: Thursday, November 12, 2020 12:35 PM To: TACC/remora <remora@noreply.github.com> Cc: Kent Milfeld <milfeld@tacc.utexas.edu>; Assign <assign@noreply.github.com> Subject: Re: [TACC/remora] Error installing 1.8.4 (#59) I grabbed the new install.sh and no joy. The error looks similar as before: ar rvs libsyscom.a common.o ioconf.o ar: creating libsyscom.a a - common.o a - ioconf.o gcc -o mpstat -g -O2 -Wall -Wstrict-prototypes -pipe -O2 mpstat.o librdstats_light.a libsyscom.a -s Installing mpstat ... ./install.sh: 79: [: 0: unexpected operator ./install.sh: 85: [: 0: unexpected operator ./install.sh: 93: ./install.sh: Syntax error: "&" unexpected I haven't tried changing the modules list - I'm going to try this next. Thanks for the help! Jeff — You are receiving this because you were assigned. Reply to this email directly, view it on GitHub<#59 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ACJ3GTCKHOXZGUI3RXVHL5TSPQTHRANCNFSM4TGRGNSQ>. This message is from an external sender. Learn more about why this matters.<https://ut.service-now.com/sp?id=kb_article&number=KB0011401>

[laytonjbgmail]

I think I can cook up something :) I can join a Zoom call on my Windows laptop where I can use Moba xterm to log into my workstation. If that works, what times work best for you? Thanks! Jeff

…

On Thu, Nov 12, 2020 at 7:37 PM Kent Milfeld ***@***.***> wrote: Hi Jeff, Can you zoom, and share your screen on your machine? If you can, I've got a UT Zoom id so you can join a meeting. (Unfortunately, you have to create a zoom account to join, which is trivial, when it asks to create an account, just provide a name and password. It is that simple. ) Best, Kent https://utexas.zoom.us/j/98588550681 ________________________________ From: laytonjbgmail ***@***.***> Sent: Thursday, November 12, 2020 12:35 PM To: TACC/remora ***@***.***> Cc: Kent Milfeld ***@***.***>; Assign < ***@***.***> Subject: Re: [TACC/remora] Error installing 1.8.4 (#59) I grabbed the new install.sh and no joy. The error looks similar as before: ar rvs libsyscom.a common.o ioconf.o ar: creating libsyscom.a a - common.o a - ioconf.o gcc -o mpstat -g -O2 -Wall -Wstrict-prototypes -pipe -O2 mpstat.o librdstats_light.a libsyscom.a -s Installing mpstat ... ./install.sh: 79: [: 0: unexpected operator ./install.sh: 85: [: 0: unexpected operator ./install.sh: 93: ./install.sh: Syntax error: "&" unexpected I haven't tried changing the modules list - I'm going to try this next. Thanks for the help! Jeff — You are receiving this because you were assigned. Reply to this email directly, view it on GitHub< #59 (comment)>, or unsubscribe< https://github.com/notifications/unsubscribe-auth/ACJ3GTCKHOXZGUI3RXVHL5TSPQTHRANCNFSM4TGRGNSQ >. This message is from an external sender. Learn more about why this matters.<https://ut.service-now.com/sp?id=kb_article&number=KB0011401> — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#59 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMYXLBOC4Z7TUSK7RZRMEHLSPQ2N7ANCNFSM4TGRGNSQ> .

[milfeld]

Jeff,
I just booted a machine with 20.0 Ubuntu from a memory drive, installed C and found that Ubuntu's /bin/sh doesn't like
a few things (using [[ instead of [ and using == instead of = within []). So, I changed install.sh to use /bin/bash, and
changed all single brackets to double brackets, for consistency. The install.sh script will not give those errors any more.

[laytonjbgmail]

Thanks a million! I was worried it was something specific with Ubuntu. Remora installed perfectly. Using it is a different story. It hangs for any code I throw at it. The remora_* directory is created correctly. It looks like the config files in that directory are created correctly. The subdirectories there such as CPU, are all created. However, there is no data in these subdirectories. I'm wondering if there is some scripting issues in the "remora" scripts themselves? Thanks a million! Jeff

…

On Fri, Nov 13, 2020 at 4:44 PM Kent Milfeld ***@***.***> wrote: Jeff, I just booted a machine with 20.0 Ubuntu from a memory drive, installed C and found that Ubuntu's /bin/sh doesn't like a few things (using [[ instead of [ and using == instead of = within []). So, I changed install.sh to use /bin/bash, and changed all single brackets to double brackets, for consistency. The install.sh script will not give those errors any more. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#59 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMYXLBKZ4PKURM4PAFCSXB3SPVO5LANCNFSM4TGRGNSQ> .

[milfeld]

Hi Jeff, Yes, I believe that is the problem. The scripts have never been hardened to continue if a problem exists in the scripts. The best approach is to minimize the number of listed modules in .../config/modules, and then set REMORA_VERBOSE to 1 (export REMORA_VERBOSE=1 and REMORA_PERIOD=1) and determine where the hang is. The first thing is that each module gets initialized, next data gets collects, and finally reports are generated. It seems that something early on is having problems. Since the data collections are done "remotely" (as a backgrouned execute, and remotely with multi-node executions) the errors are not presented to stdout/stderr. Often I have to print to a file within suspected scripts that are not working correctly. There are command to clean up, once you hang an execution. See the top level directory and the documents for them. I tried this morning running remora on an executable, but ran into problems because of using "hostname" on the "boot and try" Ubuntu (from a memory stick). I don't have a machine to install Ubuntu (I was using my wife's machine to boot a trial execution off an image.) The 20.04 didn't play well on my mac. So, hopping on your machine would be an easy and ideal way to proceed. Let's continue this correspondence through direct email (write me at milfeld@tacc.utexas.edu), this shouldn't be in a git issue. Best, Kent

________________________________ From: laytonjbgmail <notifications@github.com> Sent: Friday, November 13, 2020 12:19 PM To: TACC/remora <remora@noreply.github.com> Cc: Kent Milfeld <milfeld@tacc.utexas.edu>; Assign <assign@noreply.github.com> Subject: Re: [TACC/remora] Error installing 1.8.4 (#59) Thanks a million! I was worried it was something specific with Ubuntu. Remora installed perfectly. Using it is a different story. It hangs for any code I throw at it. The remora_* directory is created correctly. It looks like the config files in that directory are created correctly. The subdirectories there such as CPU, are all created. However, there is no data in these subdirectories. I'm wondering if there is some scripting issues in the "remora" scripts themselves? Thanks a million! Jeff

On Fri, Nov 13, 2020 at 4:44 PM Kent Milfeld ***@***.***> wrote: Jeff, I just booted a machine with 20.0 Ubuntu from a memory drive, installed C and found that Ubuntu's /bin/sh doesn't like a few things (using [[ instead of [ and using == instead of = within []). So, I changed install.sh to use /bin/bash, and changed all single brackets to double brackets, for consistency. The install.sh script will not give those errors any more. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#59 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMYXLBKZ4PKURM4PAFCSXB3SPVO5LANCNFSM4TGRGNSQ> .

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub<#59 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ACJ3GTBEBQH5TH3PQITXNWLSPV2E3ANCNFSM4TGRGNSQ>. This message is from an external sender. Learn more about why this matters.<https://ut.service-now.com/sp?id=kb_article&number=KB0011401>