Linux client unable to send blocks.

I haven't been able to send a block back yet! So far I've gotten maybe 15 or 20 of these. It's able to connect to the server to retrieve the proth test, but then can't connect to submit the finished block. Very irritating, since I'm not helping the project at all if my processed data sits in the submit queue forever :( Here is what I see:

[Fri Jan 2 16:31:32 2004] client process [v1.2.1] invoked
[Fri Jan 2 16:31:32 2004] priority set to idle
[Fri Jan 2 16:31:32 2004] connecting to server
[Fri Jan 2 16:31:32 2004] logging into server
[Fri Jan 2 16:31:32 2004] requesting a block
[Fri Jan 2 16:31:35 2004] got proth test from server (k=55459, n=5351626)
[Fri Jan 2 16:31:35 2004] server packet cached to disk
[Fri Jan 2 16:31:35 2004] AMD Athlon(tm) Processor detected. Enabling cpu specific optimizations.
[Fri Jan 2 16:57:38 2004] resolving hostname
[Fri Jan 2 16:57:38 2004] opening connection
[Fri Jan 2 17:00:47 2004] temporarily unable to connect to server -- block added to submit queue
[Fri Jan 2 17:04:21 2004] iteration: 10000/5351642 (0.19%) k = 55459 n = 5351626

I hope it can be solved! Thanks!

Read <http://www.free-dc.org/forum/showthr...&threadid=5162>.

The previous version for Linux (1.0.2) is available at <http://www-personal.engin.umich.edu/...2-Linux.tar.gz>, but you may run into problems with that version too (segmentation faults). At the moment there is no 100% stable Linux client, sadly.

--
David Norris
[email protected]

That stinks because I tried 1.0.2 first, and it segfaulted every time I ran it...

I guess I can't offer any CPU cycles at the present time.

I wonder if the Windows client would run inside Wine? Is anybody on the forum crazy enough to have actually tried it? =P

Yeah, it's pretty sad and unfortunate that we don't have a completely-functional Linux client... it should be very high on our priority list... our bad :(

--
David Norris
[email protected]

Just an idea... have you tried using Ken_g6's proxy to submit your blocks to, then sending them to the server from that? There's no real reason why it should work, unless the proxy receives blocks differently than the server, but who knows :)

Quote:

---

Originally posted by allio
Just an idea... have you tried using Ken_g6's proxy to submit your blocks to, then sending them to the server from that? There's no real reason why it should work, unless the proxy receives blocks differently than the server, but who knows :)

---

I have tested it. Client v1.21 can not connect with SBQueue either. Actually I do not use the Linuxversion of the client, but that does not matter for v1.21.

FYI,

Arnoud

I have just installed Wine 20031212 on a Debian sid system in order to try to run Windows' SB 1.2.0 on it.

Although it seems to install (it creates registry keys and successfully creates sb binaries into installation dir I'm unable to run it. It simply exits without doing anything.

I have a couple of Pentium IV with Linux iddling at this moment.

Hope a new Linux version fixing the "unable to connect" to server would be released soon.

Regards

FBCrack

As do I. For the sheer unadulterated insanity of it, did you manage to do any debugging on the broken client in Wine? Can you tell if it's causing a GPF or something, or just simply exiting on its own? I'm curious, even despite the almost total uselessness of a Wine client =P

--
David Norris
[email protected]

As far as I can tell it seems to exit cleanly by itself, I can't see any error message neither when activating the --debugmsg +all

has anybody seen this before? I don't have any machines running LINUX right now but while i ws in the ECC2 forums i noticed a link to here.

http://www.its.caltech.edu/~sgeier/code/sbwrap.html

thought someone here might wanna check it out.

I've compiled a TEST version of 1.2.0 for linux.. download it here:sb.bz2

just bunzip2 sb.bz2 wherever you have your sb stuff and run sb like normal.

note, it's test. it's not official (yet) but as far as i can tell its a lot better than 1.0.2

The TEST version works fine on my Pentium III. - I am able to connect.

Test 1.2 seg faults after CPU detection on Xeon 2.4 GHz.

TEST version seg faults on my P IV 2.4GHz after the CPU detection. I'm using Debian Woody (kernel 2.4.20) on the other way I have tried it on one of my P II it runs ok and connects with the server with no problem.

Does it dump a core file?

if not, ulimit -c unlimited .. then run it, it'll dump a core file, e-mail that core file (compressed please) to [email protected]


thanks!

linux sb1.2.0 test#2

Generate core files with the above client and send to [email protected]

Thanks,
Mike Garrison

I also got core dumps right away with sb.bz2 and SuSE

./sb sclient.conf
[Tue Jan 6 22:33:42 2004] client process [v1.2.0] invoked
[Tue Jan 6 22:33:42 2004] priority set to idle
[Tue Jan 6 22:33:42 2004] got k and n from cache
[Tue Jan 6 22:33:42 2004] Intel(R) Pentium(R) 4 CPU 2.40GHz detected. Enabling cpu specific optimizations.
[Tue Jan 6 22:33:42 2004] restarting proth test from cache (k=22699, n=5377150) [18.7%]
Segmentation fault
Do you still want core dumps for Test#1?

Test#2 is worse
You don't have permission to access /~alien88/sb.120.test2.bz2
on this server.:(

I've downloaded sb.bz2 and it appears to work fine.
1) On my dual Athlon (RedHat 9) system it appears to run a hair slower 247 down from 250. Time will tell.
2) Doesn't appear to leave network connection file open, which is a good thing.
3) I will monitor to see if it is restarted by sbwrap. Prior versions would need to be restarted a couple times a day (on average).

Thanks.

note to self: chmod file correctly.

should be able to grab it now at http://linux.redbird.com/~alien88/sb.120.test2.bz2

Quote:

---

Originally posted by bagleyd
I also got core dumps right away with sb.bz2 and SuSE

./sb sclient.conf
[Tue Jan 6 22:33:42 2004] client process [v1.2.0] invoked
[Tue Jan 6 22:33:42 2004] priority set to idle
[Tue Jan 6 22:33:42 2004] got k and n from cache
[Tue Jan 6 22:33:42 2004] Intel(R) Pentium(R) 4 CPU 2.40GHz detected. Enabling cpu specific optimizations.
[Tue Jan 6 22:33:42 2004] restarting proth test from cache (k=22699, n=5377150) [18.7%]
Segmentation fault
Do you still want core dumps for Test#1?

Test#2 is worse
You don't have permission to access /~alien88/sb.120.test2.bz2
on this server.:(

---

You can send me the core, but grab test2 and send me the core from that file too.

So far no core dumps, but the linux client (sb.bz2) still occasionally "hangs" on communication with the host. In each case the last line logged is "Receiving from server" the next line should be "Logging into server", but that was never logged.

I think this loss of communications may be related to high network activity (on this side). Two RedHat linux sb clients had been running without problems for almost 24 hours. I added two additional linux sb clients to a newly booted SuSE machine. I then started up a SuSE update adding a lot of packages to the installlation (it took over two hours to complete the update). Three of the four linux sb clients experienced a "hang" during that period (Both SuSE and one RedHat). All of the "hangs" had the "receiving from server" message as their last log message.

If I can do anthing else to help let me know.

Thanks.

Quote:

---

Originally posted by poundsja
So far no core dumps, but the linux client (sb.bz2) still occasionally "hangs" on communication with the host. In each case the last line logged is "Receiving from server" the next line should be "Logging into server", but that was never logged.

I think this loss of communications may be related to high network activity (on this side). Two RedHat linux sb clients had been running without problems for almost 24 hours. I added two additional linux sb clients to a newly booted SuSE machine. I then started up a SuSE update adding a lot of packages to the installlation (it took over two hours to complete the update). Three of the four linux sb clients experienced a "hang" during that period (Both SuSE and one RedHat). All of the "hangs" had the "receiving from server" message as their last log message.

If I can do anthing else to help let me know.

Thanks.

---

yeah, this is quite common right now and we know about it.. hopefully that will be fixed :)

:|party|: the new memory dump on exit feature is great . It will give me 3-4 extra efficient CPU-hours every day and if the client also is more stable my stats will go to the sky (anyways the sky seen from my current point of view) time will tell ...

[Thu Jan 15 18:23:04 2004] got k and n from cache
[Thu Jan 15 18:23:04 2004] Intel(R) Pentium(R) 4 CPU 2.53GHz detected. Enabling cpu specific optimizations.

--> Segmentation fault.

I have the same problem as refic on my P4 with the new Linux version, I'll just continue doing P-1 until this problem is solved.

Exactly the same on my laptop (P4).

Will we have to wait for the v3 client, or is any attempts being made at producing a working client of the existing codebase?

Well - P-1 is fun too ::jester:

how about 1.0.1 doesn't that version still work?