Is there an app for Debian/Ubuntu?

Yes, boinc is able to run on linux. Than project have to make linux version of crunching application (which this project have). So you just have to apt-get boinc client application and add burp project to your boinc client.

PS: I have question about why there is "no work" for linux platform. Can I ask for more? My linux machine is back from Christmas hibernation and have to crunch some other projects :-/

Yup, there's certainly going to be more Linux work. 2 sessions have passed the quality check but didn't render right and a further 6 or 8 are going to QA rather soon (at least 2 of which should render fine with no problems).

So in other words: Yes, there will be Linux work. It won't be right now, but it will be there soon.

Hi, I am happy about avaible work for linux machines... I have another question... Is it possible to create "longer" wus for linux? For example http://burp.renderfarming.net/workunit.php?wuid=1674767 is for me longer WU. I do not like that boinc on my desktop PC is using more disk IO resources than CPU (tasks from 15 minutes to 25 minules long take a lot of time in not cpu intensive work).

PS: just for info, how many computers using linux are capable of rendering big bunny? (enough ram, diskspace) And how many windows? (just to see how bad linux comunity is in burp :)

EDIT1: I have had look at current WU on my PC and it has taken around 8 minutes of some CPU usage (from 12% to 100%) and another 20 minutes it spent doing something with my disk (I have relatively slow raid5 for writes, but it also reads a lot.). Till mid of January I had boinc on my ssd disk and I did not noticed this weird behavior. Now I had set to boinc ionice for idle because my desktop was unusable during burp run.

When I dual boot into linux on my Windows computer, I get a high percentage of work units that error out at the end with this tag in log.
---------------------------
Exception caught: Worker application apparently died prematurely
Status: -9
---------------------------
Calling boinc_finish()...
19:49:28 (6847): called boinc_finish

The same computer under Windows rarely errors out.

Another anomaly I've noticed is that the elapsed time counter is on average about half of what the true elapsed time is. While not a big issue, it does cause awarded credits to be a lot less for Linux than what they would be for a PC for equal time.

Is it possible to create "longer" wus for linux? For example http://burp.renderfarming.net/workunit.php?wuid=1674767 is for me longer WU. I do not like that boinc on my desktop PC is using more disk IO resources than CPU (tasks from 15 minutes to 25 minules long take a lot of time in not cpu intensive work).

The running length of the workunits is not something that the project controls really. It just happens to take that long to render a full frame. On the other hand, if frames take too long to render it is sometimes possible to break them up into pieces which each take a shorter amount of time.

PS: just for info, how many computers using linux are capable of rendering big bunny? (enough ram, diskspace) And how many windows? (just to see how bad linux comunity is in burp :)

About 8.4% of the active hosts are Linux machines. That's not too bad at all.

EDIT1: I have had look at current WU on my PC and it has taken around 8 minutes of some CPU usage (from 12% to 100%) and another 20 minutes it spent doing something with my disk (I have relatively slow raid5 for writes, but it also reads a lot.). Till mid of January I had boinc on my ssd disk and I did not noticed this weird behavior. Now I had set to boinc ionice for idle because my desktop was unusable during burp run.

Sounds like an I/O-scheduler issue. Also please keep in mind that the workunits contain multiple layers and will go from 0% to 100% once for each layer (it's a known issue with Blender which has been fixed in the new clients soon to be published).

A workunit will typically write 8x200MB of data and read this multiple times during the render.

What is your raid performance through hdparm? What is the raw single-disk performance through hdparm?

When I dual boot into linux on my Windows computer, I get a high percentage of work units that error out at the end with this tag in log.
---------------------------
Exception caught: Worker application apparently died prematurely
Status: -9
---------------------------
Calling boinc_finish()...
19:49:28 (6847): called boinc_finish

The same computer under Windows rarely errors out.

This has been reported on Macs as well. How much swap space do you have on that machine?

Another anomaly I've noticed is that the elapsed time counter is on average about half of what the true elapsed time is. While not a big issue, it does cause awarded credits to be a lot less for Linux than what they would be for a PC for equal time.

This is even more odd since the credit system isn't supposed to grant credit for the elapsed time but rather for CPU time... I'll have to look into that.