I highly suspect this is gexec the application in question is a java app
which has no native method to set resources ie. Getrlimit()
Where does the following values come from?
[EMAIL PROTECTED] ~]$ gexec -n 0 bash -c 'ulimit -n'
1 1024
3 1024
2 1024
9 1024
8 1024
6 1024
7 1024
16 1024
4 1024
5 1024
10 1024
11 1024
14 1024
12 1024
15 1024
13 1024
If I ssh into any nodes in this cluster and run ulimit -n I see the
following
[EMAIL PROTECTED] ~]$ for i in $GEXEC_SVRS; do ssh $i 'ulimit -n' ;done
65536
65536
65536
65536
65536
65536
65536
65536
65536
65536
65536
65536
65536
65536
65536
65536
65536
From: Klaus Steden [mailto:[EMAIL PROTECTED]
Sent: Monday, August 25, 2008 4:32 PM
To: Rodrick Brown; [email protected]
Subject: Re: [Ganglia-general] ulimit problems with gexec
Hi Rodrick,
I'd suggest using something like strace to follow the application
through execution ... if you see an rlimit call in the output, then you
know where it's coming from ... otherwise, I'd still suspect the shell.
cordially,
Klaus
On 8/25/08 1:26 PM, "Rodrick Brown" <[EMAIL PROTECTED]>did
etch on stone tablets:
Steden, I'm not trying to set the limit beyond the systems upper limit,
currently the systems upper limit is 65636
I have no problem setting ulimit -n to any value that's less than 65536
this works fine.
I have the following entry in /etc/security/limits.conf
* - nofile 65536
The system wide limit is set to 65536
When I log one of my clusters where gexec is called by the same user and
run ulimit -a I see the following:
[EMAIL PROTECTED] ~]$ ulimit -a
core file size (blocks, -c) 0
data seg size (kbytes, -d) unlimited
max nice (-e) 0
file size (blocks, -f) unlimited
pending signals (-i) 139264
max locked memory (kbytes, -l) 32
max memory size (kbytes, -m) unlimited
open files (-n) 60000
pipe size (512 bytes, -p) 8
POSIX message queues (bytes, -q) 819200
max rt priority (-r) 0
stack size (kbytes, -s) 10240
cpu time (seconds, -t) unlimited
max user processes (-u) 139264
virtual memory (kbytes, -v) unlimited
file locks (-x) unlimited
when I start my app from gexec its using the default 1024 limit and not
what I have set in ~/.bash{rc,_profile} for that user.
So I don't understand where this 1024 value is being called from.
From: Klaus Steden [mailto:[EMAIL PROTECTED]
Sent: Monday, August 25, 2008 3:31 PM
To: Rodrick Brown; [email protected]
Subject: Re: [Ganglia-general] ulimit problems with gexec
Importance: High
Hello Rodrick,
Regular users are only able to -lower- their ulimits, never increase
them. You need to either be the superuser, or to set higher resource
limits by default. This has nothing to do with Ganglia, and it's not
something they'd be able to fix for you.
cheers,
Klaus
On 8/25/08 11:33 AM, "Rodrick Brown" <[EMAIL PROTECTED]>did
etch on stone tablets:
Can someone help me figure out what's limiting processes started from
gexec to use the default 1024 open files limit 'ulimit -n'?
In my .bash_profile/.bashrc I have ulimit -Sn 60000
Running my application manually w/o being called from gexec works fine,
however if I call gexec my app is limited to 1024 FD limit which is
causing major problems.
[EMAIL PROTECTED] ~]$ gexec -n 1 bash -c 'ulimit -Sn60000; ulimit -n'
0 /bin/bash: line 0: ulimit: open files: cannot modify limit: Invalid
argument
0 1024
[EMAIL PROTECTED] scripts]$ ssh nycplx01a 'ulimit -Sn60000; ulimit -n'
60000
I've looked at the sources for gexecd and don't see any calls to
setrlimit() so I'm not sure why this is happening any help would be
appreciated.
Thanks.
---
Rodrick R. Brown
Director, Systems Engineering
Ballista Securities, LLC
120 Wall St. Suite 2400
P: 646 307 4709
C: 347 702 0012
F: 646 219-5872
________________________________
------------------------------------------------------------------------
-
This SF.Net email is sponsored by the Moblin Your Move Developer's
challenge
Build the coolest Linux based applications with Moblin SDK & win great
prizes
Grand prize is a trip for two to an Open Source event anywhere in the
world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
<http://moblin-contest.org/redirect.php?banner_id=100&url=/>
<http://moblin-contest.org/redirect.php?banner_id=100&url=/>
________________________________
_______________________________________________
Ganglia-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ganglia-general
-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
Ganglia-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ganglia-general
