8,264 reputation
42353
bio website kamilkisiel.net
location Vancouver, Canada
age 31
visits member for 5 years, 7 months
seen Dec 30 '13 at 6:05

I've been Linux user for over 10 years, and been computing since around age 11. I got my start in programming by learning Perl on my Debian machine in high school. Since then I've learned numerous languages including C, C++, and Python. I've also dabbled in Java and assembly.

I have a bachelors degree in Computer Engineering from Simon Fraser University in Burnaby, BC, Canada.

I've worked in Japan as a CAD software developer in C++ on Windows. Not my favorite environment, but an interesting project never the less.

I'm currently employed by a Vancouver-based biotech startup working primarily on systems level projects but doing my fair share of software development.

My favorite programming language is Python and I use it nearly every day. My recent interests lie in virtualization, hpc, and large-scale systems management. I'm an active contributor to the Bcfg2 project.


Jul
16
comment Soft limit vs Hard limit?
The soft limit is what is enforced and the user can raise it to up to a maximum of the hard limit.
Jan
11
comment Howto set up SGE for CUDA devices?
You can set the value of "requestable" to be FORCED instead of YES. Then only jobs that specify a value for gpu will be considered to run in the queue. Make sure you set the default to be NONE instead of 0. Also you should just not assign the gpu complex to the non-gpu queue at all.
Jul
26
comment setting up a cluster
changed it to be more general. There are other forks (Open Grid Scheduler, Univa, even Oracle) that people might be interested in as well.
May
4
comment Sun Grid Engine VNC temporary sessions
Gridengine doesn't do anything for you there automatically, it's up to you to set up what the job actually does. One option could be to create a prolog script that decides what a free display and port are and passes them along as environment variables to your job. The other option is to have the job figure this out itself. It depends where you want to place the logic.
May
4
comment Sun Grid Engine VNC temporary sessions
There's nothing specific in gridengine that does that. It's up to you to select a unique display and corresponding port when launching VNC. What you could do is use a prolog script to set environment variables with the correct values based on what other jobs are running. Your job script would then just use those variables as arguments to vncserver.
Mar
2
comment Managing installation of non-distro software
I know about setting up repositories. The problem is I need to have multiple simultaneous versions installed and be able to switch between them.
Jan
31
comment OpenMPI in SGE fails when not observed
Can you include your parallel environment configuration?
Nov
17
comment How to limit jobs in an indicated queue in SGE
I'm not sure I understand your question. You want jobs to wait for a queue to become available, up until you reach a certain number of jobs, and only then utilize the other queue?
Nov
17
comment Sun Grid Engine (SGE) Jobs Not Visible After Adding virtual_free
There's not an easy way to do that. Personally I've made a bunch of custom commands that parse the qstat -j -xml output and display what exactly I'm looking for as the standard qstat output doesn't cover all possible use cases.
Oct
18
comment Howto set up SGE for CUDA devices?
Care to share your load sensor solution? Is there some output of nvidia-smi you can monitor to see how many GPUs are actually in use? or?
Jun
9
comment Sun GRID Engine - submit jobs that 'prefer the powerful' machine instead weaker machines
Read man sched_conf for details about that. But basically the way it works is that you can assign the same seq_no to multiple machines. In that case the load is used as a secondary sort method.
May
18
comment Xen DomU root filesystems becoming read-only on iSCSI virtual IP failover
Are you sure you are replying to the right question?
May
6
comment How to reserve complete nodes on Sun Grid Engine?
We use exclusive scheduling, it's definitely the way to go. I would upgrade GridEngine if possible.
May
3
comment Soft limit vs Hard limit?
The behaviour upon exceeding the soft limit depends on which limit it is. The man page for setrlimit(2) describes the result: linux.die.net/man/2/setrlimit .
Apr
21
comment /var/opt/ vs /var/lib
Well, the FHS is a standard created mostly for use by distributions. As a user you are free to put your data wherever you want. IMO it is mostly a matter of opinion. Personally I've never put data anywhere other than /var/lib for the reasons I've stated above. Unless you're planning to distribute software you expect to be installed on FHS systems, I think it's mostly a non-issue and more of a matter of site policy than anything else.
Mar
27
comment Limiting ssh connections with a daily amount
If lastlog isn't good enough, you can do your own accounting in your script.
Mar
3
comment Sun Grid Engine set memory requirements per jobs
Also here's a quick blog post from gridengine.info regarding memory limits: gridengine.info/2009/12/01/…
Mar
3
comment Sun Grid Engine set memory requirements per jobs
The arguments are the same as the name of the complex. So if you have h_vmem configured as a consumable resource and under a host's complex_values, your command would look like: qsub -l h_vmem=4G job1.
Mar
3
comment Sun Grid Engine set memory requirements per jobs
GridEngine will run exactly as many jobs as match the available complex resources. So if you define a node to have 16GB of h_vmem available, and you submit 5 jobs requesting 4GB, GridEngine will only place 4 of them on the node at once. Your job should always request the peak amount it expects to use. If you have swap space on your nodes, you can include that amount in the h_vmem complex value.
Mar
3
comment Parallel prologue and epilogue in Grid Engine
I feel your pain. It does seem like a deficiency in the GridEngine system if this isn't possible. You could try contacting Univa or some of the open source GridEngine authors and requesting the feature. They seem to be quite receptive to such requests at this point.