GNU/Linux S5R3 App 4.16 available for Beta test

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4260
Credit: 244791178
RAC: 21084

Looks like I was at least

Looks like I was at least partly wrong in my interpretation of the results; the client isn't catching the signal. Anyway, using an older client shouldn't harm and might give some interesting insights anyway.

BM

BM

Annika
Annika
Joined: 8 Aug 06
Posts: 720
Credit: 494410
RAC: 0

Can I change clients while I

Can I change clients while I still have uncrunched results or does it do any harm?
th3: Thanks, didn't know there were Debian packages around somewhere else than in the repositories.

Annika
Annika
Joined: 8 Aug 06
Posts: 720
Credit: 494410
RAC: 0

Well, I just changed... looks

Well, I just changed... looks like most of my WUs survived intact and BOINC is crunching away happily. My benchmarks suck a bit, which is probably due to the older core client version. I remember that problem was around for quite a while. Since two out of my three projects give fixed credit I think I can live with that. Let's hope it helps Einstein a bit, I was unable to finish a WU successfully on this box for quite a while. I don't have an Einstein WU atm so it might take a while till we know.

Conan
Conan
Joined: 19 Jun 05
Posts: 172
Credit: 7087276
RAC: 4391

I have been reading about the

I have been reading about the problems some people are having with App 4.16.

Is this App actually stable? and working now?

I am still running App 4.14 and it is working very well with no problems on my AMD Opterons with Linux (flavour Fedora Core 3 and 6).

I am trying to see the need to update to 4.16 if it is not better than 4.14.
I think I will wait till the next version comes out with more fixes.

Annika
Annika
Joined: 8 Aug 06
Posts: 720
Credit: 494410
RAC: 0

Well, atm there seem to be

Well, atm there seem to be some issues with 4.16; since we don't know where exactly they come from you can't be sure they won't affect you. So far, the problems seem to show mostly on Intel machines, but that might simply be because more people have Intel processors.
As for the need to change... I'm not sure if it will help much, performance-wise. I'm in this for the science and the technology involved ;-) if there is a new app that needs testing, I volunteer because I don't mind losing credit or having to invest some more effort if it helps the project. Also, I'm not really an inexperienced user by most standards which also helps. Therefore I volunteer to test apps which aren't quite ready for the general public, which is what beta testing is about.

th3
th3
Joined: 24 Aug 06
Posts: 208
Credit: 2208434
RAC: 0

Conan, 4.16 works perfectly

Conan,
4.16 works perfectly fine on my Intel, i bet it would on your Opterons also, but its not quite as fast as 4.14, almost no difference but a tiny bit (at least it is on Core2), so no need to change unless you really really want to help with the testing.

Donald A. Tevault
Donald A. Tevault
Joined: 17 Feb 06
Posts: 439
Credit: 73516529
RAC: 0

RE: I have been reading

Message 75237 in response to message 75234

Quote:

I have been reading about the problems some people are having with App 4.16.

Is this App actually stable? and working now?

I am still running App 4.14 and it is working very well with no problems on my AMD Opterons with Linux (flavour Fedora Core 3 and 6).

I am trying to see the need to update to 4.16 if it is not better than 4.14.
I think I will wait till the next version comes out with more fixes.

About the time that 4.16 became available, I set up a new box with CentOS Linux. Since 4.16 was there, I went ahead and gave it a try. So far, it's run perfectly, with no issues.

The machine I set it up on is an SGI 1200 server with dual P-III 700's and 256 meg of memory. The OS is CentOS 5.

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4260
Credit: 244791178
RAC: 21084

Annika, most of your tasks

Annika,

most of your tasks end up with exit code -185 and the message "Can't get shared memory segment name: can't get shared mem segment name". This is actually a message from the client and means that it couldn't set up a shared memory segment to communicate with the App. I don't know how to fix this, though.

When you fixed this, would you be willing to help us debug the "signal 11" problem in the App? The method of starting the DDD is the same I mentioned in my second post here, you would just need to type "cont" (or push the command toolbar button) once a new App is started and if it catches a signal 11, report the output (stackdump) here.

BM

BM

Annika
Annika
Joined: 8 Aug 06
Posts: 720
Credit: 494410
RAC: 0

Well, I already tried to run

Well, I already tried to run 2 WUs under the debugger and after a bit of twiddling it seemed to work okay. But I think those WUs got killed by the error you mentioned, so it probably wasn't much good.
As for fixing the problem with the core client... wished I could. I even switched to the older client version th3 mentioned here, but to no avail. Now the problem has started to affect my other projects, too. Atm I consider completely uninstalling the package (--purge) and starting from scratch in case it's something with the setup...

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4260
Credit: 244791178
RAC: 21084

I don't think it's related to

I don't think it's related to the Client or the App. Have you started to use some new program recently that may make heavy use of shared memory resources? Have you upgraded (part of) your system recently that could affect shart memory management?

BM

BM

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.