From brockp at umich.edu  Wed Dec 10 18:57:21 2014
From: brockp at umich.edu (Brock Palen)
Date: Wed, 10 Dec 2014 13:57:21 -0500
Subject: [padb-users] padb issues with openmpi 1.8
Message-ID: <B88D694E-627B-48A0-BBF6-4A77E930213B@umich.edu>

When trying to use pabd with openmpi 1.8.2  I'm getting errors:

Warning, failed to locate ranks [0-23,32-79,88-119]

I am invoking with:

#ran on the root node
padb -Ormgr=orte -a --stack-trace --tree

Its strange it can't see local cpus (ranks 0-15)  but sees some remote ones,

orte-ps gives me a full list of ranks.  So i'm confused.

Using padb 3.3

Brock Palen
www.umich.edu/~brockp
CAEN Advanced Computing
XSEDE Campus Champion
brockp at umich.edu
(734)936-1985






From ashley at pittman.co.uk  Wed Dec 10 20:31:44 2014
From: ashley at pittman.co.uk (Ashley Pittman)
Date: Wed, 10 Dec 2014 20:31:44 +0000
Subject: [padb-users] padb issues with openmpi 1.8
In-Reply-To: <B88D694E-627B-48A0-BBF6-4A77E930213B@umich.edu>
References: <B88D694E-627B-48A0-BBF6-4A77E930213B@umich.edu>
Message-ID: <FCD83A6D-306B-491A-AAC1-A650DDE0B060@pittman.co.uk>


There?s a couple of obvious things that spring to mind here, is orte-ps reporting either hostname of a FQDN for the local ranks?  If you could send me the output of orte-ps I can take a look at this tomorrow.

Ashley,

> On 10 Dec 2014, at 18:57, Brock Palen <brockp at umich.edu> wrote:
> 
> When trying to use pabd with openmpi 1.8.2  I'm getting errors:
> 
> Warning, failed to locate ranks [0-23,32-79,88-119]
> 
> I am invoking with:
> 
> #ran on the root node
> padb -Ormgr=orte -a --stack-trace --tree
> 
> Its strange it can't see local cpus (ranks 0-15)  but sees some remote ones,
> 
> orte-ps gives me a full list of ranks.  So i'm confused.
> 
> Using padb 3.3
> 
> Brock Palen
> www.umich.edu/~brockp
> CAEN Advanced Computing
> XSEDE Campus Champion
> brockp at umich.edu
> (734)936-1985
> 
> 
> 
> 
> _______________________________________________
> padb-users mailing list
> padb-users at pittman.org.uk
> http://pittman.org.uk/mailman/listinfo/padb-users_pittman.org.uk




From brockp at umich.edu  Wed Dec 10 20:36:43 2014
From: brockp at umich.edu (Brock Palen)
Date: Wed, 10 Dec 2014 15:36:43 -0500
Subject: [padb-users] padb issues with openmpi 1.8
In-Reply-To: <FCD83A6D-306B-491A-AAC1-A650DDE0B060@pittman.co.uk>
References: <B88D694E-627B-48A0-BBF6-4A77E930213B@umich.edu>
	<FCD83A6D-306B-491A-AAC1-A650DDE0B060@pittman.co.uk>
Message-ID: <0A4568E7-E018-4B43-88CE-5002AC819137@umich.edu>

Appears to not be Fully Qualified

I don't remember why (this was before I kept notes)

but I did have to make an edit to make padb work for us in the past, here is the diff:

[root at nyx bin]# diff padb padb.old 
9141,9142c9141
<             #    if ( defined $fns->{fns}{ $frame->{func} } ) {
<               if ( defined $frame->{func} and defined $fns->{fns}{ $frame->{func} } ) {
---
>                 if ( defined $fns->{fns}{ $frame->{func} } ) {

This edit is many months old, not sure why.

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: orte-ps.txt
URL: <http://pittman.org.uk/pipermail/padb-users_pittman.org.uk/attachments/20141210/83d2bbe7/attachment.txt>
-------------- next part --------------


Brock Palen
www.umich.edu/~brockp
CAEN Advanced Computing
XSEDE Campus Champion
brockp at umich.edu
(734)936-1985



> On Dec 10, 2014, at 3:31 PM, Ashley Pittman <ashley at pittman.co.uk> wrote:
> 
> 
> There?s a couple of obvious things that spring to mind here, is orte-ps reporting either hostname of a FQDN for the local ranks?  If you could send me the output of orte-ps I can take a look at this tomorrow.
> 
> Ashley,
> 
>> On 10 Dec 2014, at 18:57, Brock Palen <brockp at umich.edu> wrote:
>> 
>> When trying to use pabd with openmpi 1.8.2  I'm getting errors:
>> 
>> Warning, failed to locate ranks [0-23,32-79,88-119]
>> 
>> I am invoking with:
>> 
>> #ran on the root node
>> padb -Ormgr=orte -a --stack-trace --tree
>> 
>> Its strange it can't see local cpus (ranks 0-15)  but sees some remote ones,
>> 
>> orte-ps gives me a full list of ranks.  So i'm confused.
>> 
>> Using padb 3.3
>> 
>> Brock Palen
>> www.umich.edu/~brockp
>> CAEN Advanced Computing
>> XSEDE Campus Champion
>> brockp at umich.edu
>> (734)936-1985
>> 
>> 
>> 
>> 
>> _______________________________________________
>> padb-users mailing list
>> padb-users at pittman.org.uk
>> http://pittman.org.uk/mailman/listinfo/padb-users_pittman.org.uk
> 


From ashley at pittman.co.uk  Wed Dec 10 20:39:24 2014
From: ashley at pittman.co.uk (Ashley Pittman)
Date: Wed, 10 Dec 2014 20:39:24 +0000
Subject: [padb-users] padb issues with openmpi 1.8
In-Reply-To: <0A4568E7-E018-4B43-88CE-5002AC819137@umich.edu>
References: <B88D694E-627B-48A0-BBF6-4A77E930213B@umich.edu>
	<FCD83A6D-306B-491A-AAC1-A650DDE0B060@pittman.co.uk>
	<0A4568E7-E018-4B43-88CE-5002AC819137@umich.edu>
Message-ID: <1F6E8153-66D2-4B4E-90DB-606BC1EB2E01@pittman.co.uk>


This seems odd, I?ll try on a stock 1.8.2 install here tomorrow and let you know if I can see it this end.

Ashley.

> On 10 Dec 2014, at 20:36, Brock Palen <brockp at umich.edu> wrote:
> 
> Appears to not be Fully Qualified
> 
> I don't remember why (this was before I kept notes)
> 
> but I did have to make an edit to make padb work for us in the past, here is the diff:
> 
> [root at nyx bin]# diff padb padb.old 
> 9141,9142c9141
> <             #    if ( defined $fns->{fns}{ $frame->{func} } ) {
> <               if ( defined $frame->{func} and defined $fns->{fns}{ $frame->{func} } ) {
> ---
>>                if ( defined $fns->{fns}{ $frame->{func} } ) {
> 
> This edit is many months old, not sure why.
> 
> <orte-ps.txt>
> 
> Brock Palen
> www.umich.edu/~brockp
> CAEN Advanced Computing
> XSEDE Campus Champion
> brockp at umich.edu
> (734)936-1985
> 
> 
> 
>> On Dec 10, 2014, at 3:31 PM, Ashley Pittman <ashley at pittman.co.uk> wrote:
>> 
>> 
>> There?s a couple of obvious things that spring to mind here, is orte-ps reporting either hostname of a FQDN for the local ranks?  If you could send me the output of orte-ps I can take a look at this tomorrow.
>> 
>> Ashley,
>> 
>>> On 10 Dec 2014, at 18:57, Brock Palen <brockp at umich.edu> wrote:
>>> 
>>> When trying to use pabd with openmpi 1.8.2  I'm getting errors:
>>> 
>>> Warning, failed to locate ranks [0-23,32-79,88-119]
>>> 
>>> I am invoking with:
>>> 
>>> #ran on the root node
>>> padb -Ormgr=orte -a --stack-trace --tree
>>> 
>>> Its strange it can't see local cpus (ranks 0-15)  but sees some remote ones,
>>> 
>>> orte-ps gives me a full list of ranks.  So i'm confused.
>>> 
>>> Using padb 3.3
>>> 
>>> Brock Palen
>>> www.umich.edu/~brockp
>>> CAEN Advanced Computing
>>> XSEDE Campus Champion
>>> brockp at umich.edu
>>> (734)936-1985
>>> 
>>> 
>>> 
>>> 
>>> _______________________________________________
>>> padb-users mailing list
>>> padb-users at pittman.org.uk
>>> http://pittman.org.uk/mailman/listinfo/padb-users_pittman.org.uk
>> 
>