Hi All,

I've recently tried to install boinc from source on two machines - OpenBSD (i386) and NetBSD (sparc64) with the same basic results. Basically what I've done so far:

On OpenBSD I couldn't get the cvs source to compile, so I reverted to the archive'd 4.11 version which compiled fine. Compiled that one with --disable-server, install the client and created an app_info.xml in the projects/*/ directory (I'm currently involved with the seti project)

This is app_info.xml for openbsd ->


setiathome


setiathome-4.11.i386-unknown-openbsd3.7


setiathome
411

setiathome-4.11.i386-unknown-openbsd3.7





When running, the client appears to check-in as anonymous and download work, however, I get errors like this when the system goes to tries to actually perform the work?

2005-08-01 21:07:56 [SETI@home] execv(../../projects/setiathome.berkeley.edu/setiathome-4.11.i386-unknown-openbsd3.7) failed: -1
2005-08-01 21:07:56 [SETI@home] Unrecoverable error for result 27ja04ab.10606.5377.878414.5_3 (process exited with code 2 (0x2))
2005-08-01 21:07:56 [SETI@home] Unrecoverable error for result 27ja04ab.10606.5377.878414.5_3 (process exited with code 2 (0x2))

a similar scenario is also occuring on the netbsd machine, same execv and unrecoverable errors. I've also made a app_info.xml for that guy too:



setiathome-4.72.sparc64-unknown-netbsd2.0.2


setiathome
472

setiathome-4.72.sparc64-unknown-netbsd2.0.2





I've been able to compile the newest cvs release on the netbsd machine but haven't tinkered with the openbsd to try to get this to happen. If anyone has any pointers, please share =) I'm new to this, so it's very possible I'm missing something... just seems odd that both platforms produce the same result.

Thanks much!

i have only seen those errors when there is a syntax error in app_info.xml but that is not necessarily the only cuplrit. double check every field! executable/ usually threw me off

1) what happens if you run
$ setiathome-4.11.i386-unknown-openbsd3.7 -standalone
in a directory with a reference work unit?

2) what flags did you use for each compile?

if #1 spits out errors, it's a compile problem. but seeing that it's the same error on two completely different platforms, that wouldn't be my first guess

i have only seen those errors when there is a syntax error in app_info.xml but that is not necessarily the only cuplrit. double check every field! executable/ usually threw me off

1) what happens if you run
$ setiathome-4.11.i386-unknown-openbsd3.7 -standalone
in a directory with a reference work unit?

2) what flags did you use for each compile?

if #1 spits out errors, it's a compile problem. but seeing that it's the same error on two completely different platforms, that wouldn't be my first guess

Thanks for the response! Currently I only have the boinc_client installed in /usr/local/bin - seems I don't have the setiathome-4.11.i386-unknown-openbsd3.7 in any directory, is that something I have to link? Here's my directory structure for the location where I kick off the boinc_client process:

nemesis# ls -lR
total 100
-rw------- 1 root wheel 2351 Aug 1 21:07 account_setiathome.berkeley.edu.xml
-rw-r--r-- 1 root wheel 26159 Aug 1 21:08 client_state.xml
-rw-r--r-- 1 root wheel 26159 Aug 1 21:07 client_state_prev.xml
-rw-r--r-- 1 root wheel 745 Aug 1 14:51 global_prefs.xml
-rw-r--r-- 1 root wheel 0 Aug 1 14:48 lockfile
-rw-r--r-- 1 root wheel 12201 Aug 1 21:07 master.html
drwxr-xr-x 3 root wheel 512 Aug 1 14:49 projects
-rw-r--r-- 1 root wheel 17333 Aug 1 21:07 sched_reply_setiathome.berkeley.edu.xml
-rw-r--r-- 1 root wheel 7451 Aug 1 21:07 sched_request_setiathome.berkeley.edu.xml
drwxr-xr-x 3 root wheel 512 Aug 1 14:51 slots

./projects:
total 2
drwxr-xr-x 2 root wheel 512 Aug 1 21:07 setiathome.berkeley.edu

./projects/setiathome.berkeley.edu:
total 770
-rw------- 1 root wheel 362117 Aug 1 21:07 27ja04ab.10606.5377.878414.5
-rw------- 1 root wheel 361974 Aug 1 21:07 28ap04aa.15206.25745.811074.245
-rw-r--r-- 1 root wheel 450 Aug 1 14:49 app_info.xml

./slots:
total 2
drwxr-xr-x 2 root wheel 512 Aug 1 21:07 0

./slots/0:

Also, I didn't use any compile flags - only the ./configure --disable-server && make && make install

Did I miss a step when compiling boinc? I'm not 100% which components I need to get this going -- on Linux, I was able to do the above steps and it seems to be crunching work (didn't need the app_info.xml though)

Thx!

Thanks for the response! Currently I only have the boinc_client installed in /usr/local/bin - seems I don't have the setiathome-4.11.i386-unknown-openbsd3.7 in any directory, is that something I have to link? Here's my directory structure for the location where I kick off the boinc_client process:

nemesis# ls -lR
total 100
-rw------- 1 root wheel 2351 Aug 1 21:07 account_setiathome.berkeley.edu.xml
-rw-r--r-- 1 root wheel 26159 Aug 1 21:08 client_state.xml
-rw-r--r-- 1 root wheel 26159 Aug 1 21:07 client_state_prev.xml
-rw-r--r-- 1 root wheel 745 Aug 1 14:51 global_prefs.xml
-rw-r--r-- 1 root wheel 0 Aug 1 14:48 lockfile
-rw-r--r-- 1 root wheel 12201 Aug 1 21:07 master.html
drwxr-xr-x 3 root wheel 512 Aug 1 14:49 projects
-rw-r--r-- 1 root wheel 17333 Aug 1 21:07 sched_reply_setiathome.berkeley.edu.xml
-rw-r--r-- 1 root wheel 7451 Aug 1 21:07 sched_request_setiathome.berkeley.edu.xml
drwxr-xr-x 3 root wheel 512 Aug 1 14:51 slots

./projects:
total 2
drwxr-xr-x 2 root wheel 512 Aug 1 21:07 setiathome.berkeley.edu

./projects/setiathome.berkeley.edu:
total 770
-rw------- 1 root wheel 362117 Aug 1 21:07 27ja04ab.10606.5377.878414.5
-rw------- 1 root wheel 361974 Aug 1 21:07 28ap04aa.15206.25745.811074.245
-rw-r--r-- 1 root wheel 450 Aug 1 14:49 app_info.xml

You don't have the setiathome binary at $BOINC_HOME/projects/setiathome/ !
The boinc client don't find, and can't start the binary.
For security is better, to let the boinc client not run as root - create a user.

---

Thanks for the response! Currently I only have the boinc_client installed in /usr/local/bin - seems I don't have the setiathome-4.11.i386-unknown-openbsd3.7 in any directory, is that something I have to link? Here's my directory structure for the location where I kick off the boinc_client process:

nemesis# ls -lR
total 100
-rw------- 1 root wheel 2351 Aug 1 21:07 account_setiathome.berkeley.edu.xml
-rw-r--r-- 1 root wheel 26159 Aug 1 21:08 client_state.xml
-rw-r--r-- 1 root wheel 26159 Aug 1 21:07 client_state_prev.xml
-rw-r--r-- 1 root wheel 745 Aug 1 14:51 global_prefs.xml
-rw-r--r-- 1 root wheel 0 Aug 1 14:48 lockfile
-rw-r--r-- 1 root wheel 12201 Aug 1 21:07 master.html
drwxr-xr-x 3 root wheel 512 Aug 1 14:49 projects
-rw-r--r-- 1 root wheel 17333 Aug 1 21:07 sched_reply_setiathome.berkeley.edu.xml
-rw-r--r-- 1 root wheel 7451 Aug 1 21:07 sched_request_setiathome.berkeley.edu.xml
drwxr-xr-x 3 root wheel 512 Aug 1 14:51 slots

./projects:
total 2
drwxr-xr-x 2 root wheel 512 Aug 1 21:07 setiathome.berkeley.edu

./projects/setiathome.berkeley.edu:
total 770
-rw------- 1 root wheel 362117 Aug 1 21:07 27ja04ab.10606.5377.878414.5
-rw------- 1 root wheel 361974 Aug 1 21:07 28ap04aa.15206.25745.811074.245
-rw-r--r-- 1 root wheel 450 Aug 1 14:49 app_info.xml

You don't have the setiathome binary at $BOINC_HOME/projects/setiathome/ !
The boinc client don't find, and can't start the binary.
For security is better, to let the boinc client not run as root - create a user.

I think we're getting somewhere :) Is that binary seperate from the boinc-4.11.tar.gz source code? Do I need to compile something else and put it there?

---

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"Pain dosen't hurt, when it's all you have ever felt"

Thanks for the response! Currently I only have the boinc_client installed in /usr/local/bin - seems I don't have the setiathome-4.11.i386-unknown-openbsd3.7 in any directory, is that something I have to link? Here's my directory structure for the location where I kick off the boinc_client process:

nemesis# ls -lR
total 100
-rw------- 1 root wheel 2351 Aug 1 21:07 account_setiathome.berkeley.edu.xml
-rw-r--r-- 1 root wheel 26159 Aug 1 21:08 client_state.xml
-rw-r--r-- 1 root wheel 26159 Aug 1 21:07 client_state_prev.xml
-rw-r--r-- 1 root wheel 745 Aug 1 14:51 global_prefs.xml
-rw-r--r-- 1 root wheel 0 Aug 1 14:48 lockfile
-rw-r--r-- 1 root wheel 12201 Aug 1 21:07 master.html
drwxr-xr-x 3 root wheel 512 Aug 1 14:49 projects
-rw-r--r-- 1 root wheel 17333 Aug 1 21:07 sched_reply_setiathome.berkeley.edu.xml
-rw-r--r-- 1 root wheel 7451 Aug 1 21:07 sched_request_setiathome.berkeley.edu.xml
drwxr-xr-x 3 root wheel 512 Aug 1 14:51 slots

./projects:
total 2
drwxr-xr-x 2 root wheel 512 Aug 1 21:07 setiathome.berkeley.edu

./projects/setiathome.berkeley.edu:
total 770
-rw------- 1 root wheel 362117 Aug 1 21:07 27ja04ab.10606.5377.878414.5
-rw------- 1 root wheel 361974 Aug 1 21:07 28ap04aa.15206.25745.811074.245
-rw-r--r-- 1 root wheel 450 Aug 1 14:49 app_info.xml

You don't have the setiathome binary at $BOINC_HOME/projects/setiathome/ !
The boinc client don't find, and can't start the binary.
For security is better, to let the boinc client not run as root - create a user.

I think we're getting somewhere :) Is that binary seperate from the boinc-4.11.tar.gz source code? Do I need to compile something else and put it there?

Ah I see - I need the seti client & the boinc client.. seti client gets compiled and copied to project dir, is this right? On Linux, I only installed the boinc client and it seems to be producing results.. do I need to do this for any install? Or only ones that use app_info.xml for anonymous platform

app_info.xml tells boinc what client to use. if you did not have app_info.xml in place, boinc tries to down the default precompiled seti client. on your linux install, your boinc client downloaded the default linux seti client automatically. on non-mainstream systems using BSD and sparc etc there is no default client. boinc won't be able to find it and you would see a different error (if there was no app_info.xml in place)

so you have to compile these (essentially the same way you did the boinc client). hopefully you didn't delete your boinc compile directory because seti needs to reference files it created.

once compiled, move the executables to projects/setiathome.berkeley.edu/ and you should be all set to go

app_info.xml tells boinc what client to use. if you did not have app_info.xml in place, boinc tries to down the default precompiled seti client. on your linux install, your boinc client downloaded the default linux seti client automatically. on non-mainstream systems using BSD and sparc etc there is no default client. boinc won't be able to find it and you would see a different error (if there was no app_info.xml in place)

so you have to compile these (essentially the same way you did the boinc client). hopefully you didn't delete your boinc compile directory because seti needs to reference files it created.

once compiled, move the executables to projects/setiathome.berkeley.edu/ and you should be all set to go

Thanks for setting me straight =) I've compiled the client successfully on NetBSD so far, put it in the project directory but I'm now coming up with this error. The box seemed to compute a partial result based off the stats page.

2005-08-02 12:16:56 [SETI@home] Reason: To fetch work
2005-08-02 12:16:56 [SETI@home] Requesting 172800 seconds of work, returning 0 results
2005-08-02 12:16:58 [SETI@home] Scheduler request to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi succeeded
2005-08-02 12:16:59 [SETI@home] Started download of 26fe05aa.10476.15490.829832.232
2005-08-02 12:17:03 [SETI@home] Finished download of 26fe05aa.10476.15490.829832.232
2005-08-02 12:17:03 [SETI@home] Throughput 114116 bytes/sec
2005-08-02 12:17:03 [---] request_reschedule_cpus: files downloaded
2005-08-02 12:17:03 [SETI@home] Starting result 26fe05aa.10476.15490.829832.232_1 using setiathome version 4.18
2005-08-02 12:21:33 [SETI@home] Unrecoverable error for result 26fe05aa.10476.15490.829832.232_1 (process exited with code 131 (0x83))
2005-08-02 12:21:33 [SETI@home] Unrecoverable error for result 26fe05aa.10476.15490.829832.232_1 (process exited with code 131 (0x83))
2005-08-02 12:21:33 [---] request_reschedule_cpus: process exited
2005-08-02 12:21:33 [SETI@home] Deferring communication with project for 1 minutes and 0 seconds
2005-08-02 12:21:33 [SETI@home] Deferring communication with project for 1 minutes and 0 seconds
2005-08-02 12:21:33 [SETI@home] Computation for result 26fe05aa.10476.15490.829832.232_1 finished
2005-08-02 12:21:34 [SETI@home] Started upload of 26fe05aa.10476.15490.829832.232_1_0
2005-08-02 12:21:37 [SETI@home] Finished upload of 26fe05aa.10476.15490.829832.232_1_0

I've moved all my old xml files from /root to ~myhome and am now executing the boinc_client as a user-- recursively changed ownership to the proper user and group and this is now my dir structure.

This is what happens when I try to execute the work manually. The client runs for about a minute and then dumps.

ghey: {65} ../../projects/setiathome.berkeley.edu/setiathome-4.18.sparc64-unknown-netbsd2.0.2 -standalone
Floating point exception (core dumped)
ghey: {66} pwd
/home/gavin/seti/slots/0


Sigh.... not enough memory?

cpu0 at mainbus0: SUNW,UltraSPARC-IIi @ 299.841 MHz, version 0 FPU
total memory = 384 MB
avail memory = 364 MB

2005-08-02 12:17:03 [SETI@home] Starting result 26fe05aa.10476.15490.829832.232_1 using setiathome version 4.18
2005-08-02 12:21:33 [SETI@home] Unrecoverable error for result 26fe05aa.10476.15490.829832.232_1 (process exited with code 131 (0x83))
2005-08-02 12:21:33 [SETI@home] Unrecoverable error for result 26fe05aa.10476.15490.829832.232_1 (process exited with code 131 (0x83))
2005-08-02 12:21:33 [---] request_reschedule_cpus: process exited
Sigh.... not enough memory?

Your installation looks OK. Physical Memory is OK.

The error code 131 could be a lot of issues. One of them could be, that you don't have enough shared memory, or some kernel parameters are to a low (maxdsiz, maxssiz). Sorry don't know it exactly for BSD.
Here you can eventualty find some hints : www.dotsch.de/seti/hpux/install

Otherwithe you could search in the forums or at www.boinc-doc.net for more detailed infos...

Lars

---

Can you please post the output from stderr.txt.

---

Can you please post the output from stderr.txt.

That was created when I was trying to figure out how to run the seti client -standalone

I finally figured out that I had to be in the slots/0 directory to run it and actually have it perform some work. Here's what it said initially

Can't open init data file - running in standalone mode
SETI@home error -5 Can't open file
(work_unit.sah) in read_wu_state() errno=2

File: worker.cpp
Line: 120


No updates were made to that file once the boinc_client started reporting the 131 error =( I've changed the soft limits on my system to unlimited on all, running the client again to see if that made any difference.

Can you please post the output from stderr.txt.

Hrm. The errors seem to be gone when I run boinc_client, BUT the work that's downloaded is finishing rather quickly. Take a look at this:

2005-08-02 13:52:05 [---] request_reschedule_cpus: files downloaded
2005-08-02 13:52:05 [SETI@home] Starting result 27ja04ab.10606.8578.648558.221_2 using setiathome version 4.18
2005-08-02 13:53:16 [---] request_reschedule_cpus: process exited
2005-08-02 13:53:16 [SETI@home] Computation for result 27ja04ab.10606.8578.648558.221_2 finished
2005-08-02 13:53:17 [SETI@home] Sending scheduler request to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi
2005-08-02 13:53:17 [SETI@home] Reason: To fetch work
2005-08-02 13:53:17 [SETI@home] Requesting 172800 seconds of work, returning 0 results
2005-08-02 13:53:17 [SETI@home] Started upload of 27ja04ab.10606.8578.648558.221_2_0
2005-08-02 13:53:18 [SETI@home] Scheduler request to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi succeeded
2005-08-02 13:53:18 [SETI@home] Message from server: Not sending work - last RPC too recent: 77 sec
2005-08-02 13:53:18 [SETI@home] No work from project
2005-08-02 13:53:18 [SETI@home] No work from project
2005-08-02 13:53:19 [SETI@home] Deferring communication with project for 10 minutes and 4 seconds
2005-08-02 13:53:19 [SETI@home] Deferring communication with project for 10 minutes and 4 seconds
2005-08-02 13:53:21 [SETI@home] Finished upload of 27ja04ab.10606.8578.648558.221_2_0
2005-08-02 13:53:21 [SETI@home] Throughput 217181 bytes/sec

As you can see, it finishes in about a minute.

This is the most recent result from the stats page:

2 Aug 2005 20:33:35 UTC 2 Aug 2005 20:52:04 UTC Over Success Done 489.91 0.12 pending

Am I going to have to live with this because my system sucks? Does seti throttle the amount of work based on hard disk space/etc ?

Domain name ghey
Local Standard Time UTC -7 hours
Name ghey
Created 2 Aug 2005 3:56:52 UTC
Total Credit 0.00
Recent average credit 0.00
CPU type sparc64
SUNW,Ultra-5_10 (SUNW,UltraSPARC-IIi @ 299.836 MHz, version 0 FPU)
Number of CPUs 1
Operating System NetBSD
2.0.2
Memory 384 MB
Cache 976.56 KB
Swap space 0 MB
Total disk space 3.76 GB
Free Disk Space 3.18 GB
Measured floating point speed 136.22 million ops/sec
Measured integer speed 274.01 million ops/sec
Average upload rate 2.86 KB/sec
Average download rate 119.75 KB/sec
Average turnaround time 0 days
Maximum daily WU quota per CPU 100/day
Results 9
Number of times client has contacted server 14
Last time contacted server 2 Aug 2005 20:53:21 UTC
% of time BOINC client is running 97.8622 %
While BOINC running, % of time host has an Internet connection -100 %
While BOINC running, % of time work is allowed 100 %
Average CPU efficiency 0.999732
Result duration correction factor 0.980127

It could be, that the disk quota changes was the cause of the problem.
But if i look at the results from your Ultra 5/10 with NetBSD it reports shared memory errors :
4.72
process exited with code 131 (0x83)

Can't set up shared mem: -1
SIGSEGV: segmentation violation
Exiting...
SIGSEGV: segmentation violation
Exiting...


The other two "successfull" WUs reports :
SETI@Home Informational message -9 result_overflow
NOTE: The number of results detected exceeds the storage space allocated.



Try to increase you shared memory parameters !

Otherwise, there was a problem in the boinc client source. You can try to apply this changes to the boinc_public/lib/shmem.C, and recompile it :

Search for
id = shmget(key, size, IPC_CREAT|0666);
and change it into :
id = shmget(key, size, IPC_CREAT|SHM_R|SHM_W);


Do you have let let seti binary successfull run the reference workunit ?

---

It could be, that the disk quota changes was the cause of the problem.
But if i look at the results from your Ultra 5/10 with NetBSD it reports shared memory errors :
4.72
process exited with code 131 (0x83)

Can't set up shared mem: -1
SIGSEGV: segmentation violation
Exiting...
SIGSEGV: segmentation violation
Exiting...


The other two "successfull" WUs reports :
SETI@Home Informational message -9 result_overflow
NOTE: The number of results detected exceeds the storage space allocated.



Try to increase you shared memory parameters !

Otherwise, there was a problem in the boinc client source. You can try to apply this changes to the boinc_public/lib/shmem.C, and recompile it :

Search for
id = shmget(key, size, IPC_CREAT|0666);
and change it into :
id = shmget(key, size, IPC_CREAT|SHM_R|SHM_W);


Do you have let let seti binary successfull run the reference workunit ?

The successful ones started after I stopped boinc & set ulimit to unlimited on all values. Now that the successfuls aren't producting an error result, any ideas why they're completing so fast? Typical work takes hours I'd figure. Hmm

It could be, that the disk quota changes was the cause of the problem.
But if i look at the results from your Ultra 5/10 with NetBSD it reports shared memory errors :
4.72
process exited with code 131 (0x83)

Can't set up shared mem: -1
SIGSEGV: segmentation violation
Exiting...
SIGSEGV: segmentation violation
Exiting...


The other two "successfull" WUs reports :
SETI@Home Informational message -9 result_overflow
NOTE: The number of results detected exceeds the storage space allocated.



Try to increase you shared memory parameters !

Otherwise, there was a problem in the boinc client source. You can try to apply this changes to the boinc_public/lib/shmem.C, and recompile it :

Search for
id = shmget(key, size, IPC_CREAT|0666);
and change it into :
id = shmget(key, size, IPC_CREAT|SHM_R|SHM_W);


Do you have let let seti binary successfull run the reference workunit ?

The successful ones started after I stopped boinc & set ulimit to unlimited on all values. Now that the successfuls aren't producting an error result, any ideas why they're completing so fast? Typical work takes hours I'd figure. Hmm

Nevermind. Getting error codes again. Bleh..

But if i look at the results from your Ultra 5/10 with NetBSD it reports shared memory errors :
4.72
process exited with code 131 (0x83)

Can't set up shared mem: -1
SIGSEGV: segmentation violation
Exiting...
SIGSEGV: segmentation violation
Exiting...

Nevermind. Getting error codes again. Bleh..[/quote]

I get the same problems with NetBSD on x86. I looked also in the forums, and it looks like it is a problem on NetBSD and OpenBSD.

---

I heave somewhere readen, that it is posible to let FreeBSD binaries run on OpenBSD and NetBSD.
Is there a los in performance ? - If not, I think it is worth a try to install the running FreeBSD binaries on OpenBSD or NetBSD.

---

Hello Paul,

I had simliar problems to compile the seti app on NetBSD x86 like you.
The reference WU sometime aborts without any error, somtimes with the result_overflow error.
I could fix this by using "-ffloat-store -funsafe-math-optimizations" in the CFLAGS (with no other further options).

I also compiled the seti app for OpenBSD, and get the same behavior, which was solved by using this CFLAGS. The reference WU was successfull run, and in real live tests I have now two WUs was validated. I will release the binary for OpenBSD x86 in a couple of days.
If you which, I can mail you the binaries for testing.

Regards

Lars

---

Hello Paul,

I had simliar problems to compile the seti app on NetBSD x86 like you.
The reference WU sometime aborts without any error, somtimes with the result_overflow error.
I could fix this by using "-ffloat-store -funsafe-math-optimizations" in the CFLAGS (with no other further options).

I also compiled the seti app for OpenBSD, and get the same behavior, which was solved by using this CFLAGS. The reference WU was successfull run, and in real live tests I have now two WUs was validated. I will release the binary for OpenBSD x86 in a couple of days.
If you which, I can mail you the binaries for testing.

Regards

Lars

What's the difference between Free,Open and Net BSD ?
Could anyone explain this to me please ?
Arnt these BSD version binary compatible ?

---

Join BOINC United now!

If you which, I can mail you the binaries for testing.

The binaries are now available for download at http://www.lb.shuttle.de/apastron/boincDown.shtml

---