clusterssh runaway process on start

Bug #563952 reported by Ben Hodgens
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
clusterssh (Ubuntu)
New
Undecided
Unassigned

Bug Description

Binary package hint: clusterssh

Release: Ubuntu 9.10 x64
Package version: 3.26-1
Perl version: 5.10.0-24ubuntu4

Symptoms: upon starting clusterssh with a host list of 1 or more, or starting clusterssh and manually opening a host, cssh will start consuming massive amounts of memory and not performing the requested task. For instance, this is output from top approximately 5 seconds after running "cssh hostname" (or specifying a host after cssh is already running). Note RES and SHR:

  PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
 4837 username 20 0 2126m 2.0g 416 D 62 52.0 0:24.66 cssh 0

Entering "debug 4" in the configuration file does not yield any output when cssh is run from a terminal. If the process is not killed quickly, it will hang the system when it runs out of RAM and VM.

This problem consistently replicates itself across reboots. It started to occur just recently without any csshrc changes; I'm using the stock csshrc file with cluster definitions.

Additionally, I am running the awesome window manager, and the unattended-upgrades package is installed (potentially narrowing down how this regression occurred).

Revision history for this message
Tony Mancill (tmancill) wrote : Re: [Bug 563952] [NEW] clusterssh runaway process on start

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Ben,

Does the problem exist when running another window manager? I suspect the issue
may lie in the interaction between perl-tk and the WM, or in a change to the
terminal emulator being used. It sounds as if the issue is occurring when CSSH
attempts to open a terminal.

Thank you,
Tony

On 4/15/2010 8:48 AM, Ben Hodgens wrote:
> Public bug reported:
>
> Binary package hint: clusterssh
>
> Release: Ubuntu 9.10 x64
> Package version: 3.26-1
> Perl version: 5.10.0-24ubuntu4
>
> Symptoms: upon starting clusterssh with a host list of 1 or more, or
> starting clusterssh and manually opening a host, cssh will start
> consuming massive amounts of memory and not performing the requested
> task. For instance, this is output from top approximately 5 seconds
> after running "cssh hostname" (or specifying a host after cssh is
> already running). Note RES and SHR:
>
> PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
> 4837 username 20 0 2126m 2.0g 416 D 62 52.0 0:24.66 cssh 0
>
> Entering "debug 4" in the configuration file does not yield any output
> when cssh is run from a terminal. If the process is not killed quickly,
> it will hang the system when it runs out of RAM and VM.
>
> This problem consistently replicates itself across reboots. It started
> to occur just recently without any csshrc changes; I'm using the stock
> csshrc file with cluster definitions.
>
> Additionally, I am running the awesome window manager, and the
> unattended-upgrades package is installed (potentially narrowing down how
> this regression occurred).
>
> ** Affects: clusterssh (Ubuntu)
> Importance: Undecided
> Status: New
>
>
> ** Tags: condition leak memory race
>

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkvKKpAACgkQpdwBkPlyvgP5agCeO3FbecMi4a8JvoQOSBNOva0X
eJYAnj1PIarPx4jvVCsZ1tlKx+PlEdCs
=wjWJ
-----END PGP SIGNATURE-----

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.