A few hours ago, boinc.exe started crashing on 2 of my systems for no apparent reason after running very stably for months. I was using 6.3.6, and then downgraded to 6.2.14 with same results. Is anybody else experiencing this? And, more importantly, does anybody have a fix?

---

The odd number in the middle of the 3 cigits 6.3.6 means that it's a test version.
If you're doing alpha testing, then you should post on the apha test forum.

As always:
What OS?
32 bit or 64 bit BOINC?
Crashing in what way?

[quote]The odd number in the middle of the 3 cigits 6.3.6 means that it's a test version.
If you're doing alpha testing, then you should post on the apha test forum.

As always:
What OS?
32 bit or 64 bit BOINC?
Crashing in what way?

I rolled back to 6.2.14 with same result.

XP Professional, 32 bit.

Attempts tp connect and I get a windows message that "BOINC client has encountered a problem and needs to close.' It happened on two machines in different locatins within a few hours of each other.

Thanks

---

[quote]The odd number in the middle of the 3 cigits 6.3.6 means that it's a test version.
If you're doing alpha testing, then you should post on the apha test forum.

As always:
What OS?
32 bit or 64 bit BOINC?
Crashing in what way?

I rolled back to 6.2.14 with same result.

XP Professional, 32 bit.

Attempts tp connect and I get a windows message that "BOINC client has encountered a problem and needs to close.' It happened on two machines in different locatins within a few hours of each other.

Thanks

I am having the same problem. Is there a fix?

---

I lost boinc.exe on 5.10.45. Doesn't matter what version I try to run I can't keep boinc.exe running. I believe it was caused by one of the projects, but not sure which one at this time. Tracing it back.

---

I'm busy with it people, as I have exactly the same problem on my Win2k machine with 6.2.14, 6.2.15 and 6.2.16
When installed as a service the service hangs. When installed as not a service, boinc.exe will crash.

I narrowed it down to something in my data directory, but need Rom to take a look at that. Fun is that the directory is 305MB when 7zipped. So it will take some fun times to transfer that to Rom.

I've flagged this thread for his attention though.

---

Please people, list the projects you are attached to and that can fetch work or do other communications even while on NNT.
Also please state which Operating System and BOINC version you use. Not just chime in "me too".

I'll start.

Almeregrid (NNT, communicates)
QCN Alpha (NNT, trickles)
CPDN Beta2 (NNT, trickles)
Seti
Primegrid
Pirates
Genetic Life
Enigma
Intelligence Realm
Cosmology
Einstein
LHC (no work)
Predictor (no work)

---

Sorry about that, mines a small list,

Genetic Life
Hydrogen
Orbit (No work)

Orbit doesn't have any work right now and the last task I got and project it contacted was Hydrogen.

---

Same or similar problem, running BOINC 5.10.30 on Windows 2K, which has been stable for months. I suspect it is an application problem, not the client core or manager.

I first noticed it with a pop-up that said

boinc.exe - Application error

"The exception unknown software exception (0xc00000409) occured in the application at location 0x00420067"

Then the manager complains that it cannot connect to client. Would I like to try again? Yes, and the cycle repeats. So it looks like a problem starting the client when it is starting an app.

Looking at client_state.xml shows two active tasks:

WCG: R00063_84.... v617
QMC: four_412_peptidsm-ecp2 v501

Not sure which is the culprit, or if it's actually one of these.
I'll check the slots....

---

-- Eric Myers

"Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats

Eric Myers wrote:

I'll check the slots....

Slot 0 (WCG) has not had activity since Friday.

Slot 2 (QMC) had recent acivity yesterday afternoon, which is when I think I first saw this. There are two checkpoint files, tmpChkpoint000 and Chkpoint000. Also a file called tmpchk. Maybe it had a problem during checkpointing and cannot recover? stderr.txt has complaints of "No heartbeat from core client for 31 sec - exiting"

As an experiment I stopped BOINC, removed tmpChkpoint000 and tmpchk, and started BOINC. It takes longer to get there, but eventually the same error(s).

So I also removed Chkpoint000. Immediately the same error.

---

-- Eric Myers

"Education is not the filling of a pail, but the lighting of a fire." -- William Butler Yeats

Checking my stdoutdae.txt I see that for each of the instances where I started BOINC last night and the service hung up, BOINC was running... but it didn't read any of the account files.

I have all of the same reactions (protected/service mode):
08-Aug-2008 03:25:40 [---] Starting BOINC client version 6.2.14 for windows_intelx86
08-Aug-2008 03:25:40 [---] log flags: task, file_xfer, sched_ops, unparsed_xml, file_xfer_debug, sched_op_debug
08-Aug-2008 03:25:40 [---] log flags: benchmark_debug, scrsave_debug, checkpoint_debug
08-Aug-2008 03:25:40 [---] Libraries: libcurl/7.18.0 OpenSSL/0.9.8e zlib/1.2.3
08-Aug-2008 03:25:40 [---] Running as a daemon
08-Aug-2008 03:25:40 [---] Data directory: C:\BOINC Data
08-Aug-2008 03:25:40 [---] Running under account boinc_master
08-Aug-2008 03:25:43 [SETI@home] Found app_info.xml; using anonymous platform
08-Aug-2008 03:25:44 [---] Processor: 1 AuthenticAMD AMD Athlon(tm) XP 2200+ [x86 Family 6 Model 8 Stepping 1]
08-Aug-2008 03:25:44 [---] Processor features: fpu tsc sse 3dnow mmx
08-Aug-2008 03:25:44 [---] OS: Microsoft Windows 2000: Professional Edition, Service Pack 4, (05.00.2195.00)
08-Aug-2008 03:25:44 [---] Memory: 1023.48 MB physical, 2.41 GB virtual
08-Aug-2008 03:25:44 [---] Disk: 10.26 GB total, 5.50 GB free
08-Aug-2008 03:25:44 [---] Local time is UTC +2 hours


With the (installed as non-protected, so not a service) crashing boinc.exe I see:
08-Aug-2008 06:08:01 [---] Starting BOINC client version 6.2.16 for windows_intelx86
08-Aug-2008 06:08:01 [---] log flags: task, file_xfer, sched_ops, unparsed_xml, file_xfer_debug, sched_op_debug
08-Aug-2008 06:08:01 [---] log flags: benchmark_debug, scrsave_debug, checkpoint_debug
08-Aug-2008 06:08:01 [---] Libraries: libcurl/7.18.0 OpenSSL/0.9.8e zlib/1.2.3
08-Aug-2008 06:08:01 [---] Data directory: C:\BOINC Data2
08-Aug-2008 06:08:01 [---] Running under account Administrator
08-Aug-2008 06:08:01 [SETI@home] Found app_info.xml; using anonymous platform
08-Aug-2008 06:08:01 [---] Processor: 1 AuthenticAMD AMD Athlon(tm) XP 2200+ [x86 Family 6 Model 8 Stepping 1]
08-Aug-2008 06:08:01 [---] Processor features: fpu tsc sse 3dnow mmx
08-Aug-2008 06:08:01 [---] OS: Microsoft Windows 2000: Professional Edition, Service Pack 4, (05.00.2195.00)
08-Aug-2008 06:08:01 [---] Memory: 1023.48 MB physical, 2.41 GB virtual
08-Aug-2008 06:08:01 [---] Disk: 10.26 GB total, 5.47 GB free
08-Aug-2008 06:08:01 [---] Local time is UTC +2 hours

It does NOT read the account files. Then BOINC crashes or the service hangs.
(also missing is the No co-processors line, but 5.10 doesn't do that. So not sure if that's causing it.

---

Just had the same problem. I had a QCN account and deleted that xml and everything started right back up upon restarting BOINC. It seems to try and update on its own and is blocked by windows. Hope your problem is the same.

Just had the same problem. I had a QCN account and deleted that xml and everything started right back up upon restarting BOINC. It seems to try and update on its own and is blocked by windows. Hope your problem is the same.

Heres what I did. I manually deleted the boinc folder, then downloaded 6.2.14 and installed it. Then I attached WCG. It works great. After while I attached QCN then it failed. I manually deleted the QCN acct file from the BOINC folder and restarted both the manager and client. It is still working. There is something wrong with QCN or it you start a second project. this is happening on my HP Laptop running vista 32.

---

I've uploaded my whole BOINC Data directory to Skydrive, Rom is busy downloading it to test with. Either I or he will give an update on what may cause this.

I don't believe it's QCN, as some people here aren't even attached to it and have the problem. I do believe however that it is caused by an account file. I am attached to 37 projects, so it's going to be difficult to find out which one is causing it.

---

I've had the same problem - on both 5.10.13 and 5.10.45.

Had work running for orbit@home and qcn.

have tried several things before posting on seti@home message boards. One thing I've noticed is, that the active_task_set section in client_state.xml is empty now, so I suppose work is lost on those wu's.

I do believe however that it is caused by an account file. I am attached to 37 projects, so it's going to be difficult to find out which one is causing it.

I'm pretty sure you're right about it being an account file - it's definitely not the client_state.xml file, or any wu or app. I have tried several things, and even when boinc has no knowledge of any app and wu assoiciated with projects through the client_state.xml-file boinc.exe won't stay up.

Had work running for orbit@home and qcn.

Yes, but you're also attached to about every project that exists. ;-)

I've put my remaining rig on total NNT as I don't want to lose the Orbit task at 70% (545 hours running)

---

Had work running for orbit@home and qcn.

Yes, but you're also attached to about every project that exists. ;-)

*lol* can't argue with that ;) If it is just one account file that creates the trouble, it's better for Rom to go through your folder than mine ;)

I've put my remaining rig on total NNT as I don't want to lose the Orbit task at 70% (545 hours running)

Understandable - for me, I only hope that the task I was working on can be re-started and won't be void - think that's max I can hope for. The QCN-tasks are a bit special, so I don't think it would be much of a problem to lose that task

I noticed MikeMarsUk having the same problem. Reposting his message here:

MikeMarsUK wrote:

My Boinc 6 installation (6.2.14) on the AMD has crashed out and won't restart (I get a 'boinc client has crashed' message). I've tried downgrading it to 5.10.45 and the same happens.

I'm comparing the crashed installation to a backup to see if I can see what the problem is.

I seem to recall that one time people had a similar problem, it was due to libcurl crashing due to an iffy redirect on LHC. Has anyone heard complaints of similar problems again?

-- Edit:

The client_state.xml looks fine. I've tried forcing cpu suspended + network suspended by editing the xml but that doesn't seem to make much difference.
"stdoutgui.txt" wrote:

[08/07/08 23:27:30] TRACE [3360]: RPC_CLIENT::init connect on 228 returned -1

[08/07/08 23:27:30] TRACE [3360]: RPC_CLIENT::init boinc_socket returned 328

[08/07/08 23:27:30] TRACE [3360]: RPC_CLIENT::init connect returned -1

[08/07/08 23:27:30] TRACE [3360]: RPC_CLIENT::init attempting connect

[08/07/08 23:27:31] TRACE [3360]: RPC_CLIENT::init_poll sock = 328

[08/07/08 23:27:31] TRACE [3360]: RPC_CLIENT::init_poll sock = 328

[08/07/08 23:27:31] TRACE [3360]: RPC_CLIENT::init_poll sock = 328

[08/07/08 23:27:31] TRACE [3360]: RPC_CLIENT::init_poll sock = 328

[08/07/08 23:27:31] TRACE [3360]: RPC_CLIENT::init_poll attempting connect

[08/07/08 23:27:32] TRACE [3360]: RPC_CLIENT::init_poll sock = 328

[08/07/08 23:27:32] TRACE [3360]: RPC_CLIENT::init_poll sock = 328

[08/07/08 23:27:32] TRACE [3360]: RPC_CLIENT::init_poll sock = 328

[08/07/08 23:27:32] TRACE [3360]: RPC_CLIENT::init_poll sock = 328

[08/07/08 23:27:33] TRACE [3360]: RPC_CLIENT::init_poll sock = 328

[08/07/08 23:27:33] TRACE [3360]: RPC_CLIENT::init_poll attempting connect

[08/07/08 23:27:33] TRACE [3360]: RPC_CLIENT::init_poll sock = 328

[08/07/08 23:27:33] TRACE [3360]: RPC_CLIENT::init_poll sock = 328

[08/07/08 23:27:33] TRACE [3360]: RPC_CLIENT::init_poll sock = 328

[08/07/08 23:27:34] TRACE [3360]: RPC_CLIENT::init_poll sock = 328

[08/07/08 23:27:34] TRACE [3360]: RPC_CLIENT::init_poll attempting connect

... etc etc ...

Nothing appears after the disk size line in the log (without debug options) when I start up the client. With debug lines, only a little more appears:

"stdoutdae.txt" wrote:

E:\Program Files\BOINC>notepad cc_config.xml

E:\Program Files\BOINC>boinc.exe
08-Aug-2008 00:25:43 [---] Starting BOINC client version 5.10.45 for windows_intelx86
08-Aug-2008 00:25:43 [---] log flags: task, file_xfer, sched_ops, cpu_sched, rr_simulation, task_debug
08-Aug-2008 00:25:43 [---] log flags: work_fetch_debug, unparsed_xml, state_debug, file_xfer_debug, sched_op_debug
08-Aug-2008 00:25:43 [---] log flags: http_debug, proxy_debug, time_debug, http_xfer_debug, benchmark_debug
08-Aug-2008 00:25:43 [---] log flags: poll_debug, guirpc_debug, scrsave_debug, app_msg_send, app_msg_receive
08-Aug-2008 00:25:43 [---] log flags: mem_usage_debug, network_status_debug, checkpoint_debug
08-Aug-2008 00:25:43 [---] Libraries: libcurl/7.18.0 OpenSSL/0.9.8e zlib/1.2.3
08-Aug-2008 00:25:43 [---] Data directory: E:\Program Files\BOINC
08-Aug-2008 00:25:43 [---] [unparsed_xml] PROJECT::parse_account(): unrecognized: <allow_beta_work>1</allow_beta_work>
08-Aug-2008 00:25:43 [---] [unparsed_xml] PROJECT::parse_account(): unrecognized: <apps_selected>
08-Aug-2008 00:25:43 [---] [unparsed_xml] PROJECT::parse_account(): unrecognized: <app_id>9</app_id>
08-Aug-2008 00:25:43 [---] [unparsed_xml] PROJECT::parse_account(): unrecognized: <app_id>10</app_id>
08-Aug-2008 00:25:43 [---] [unparsed_xml] PROJECT::parse_account(): unrecognized: <app_id>13</app_id>
08-Aug-2008 00:25:43 [---] [unparsed_xml] PROJECT::parse_account(): unrecognized: </apps_selected>
08-Aug-2008 00:25:43 [---] [state_debug] set dirty: Set mode
08-Aug-2008 00:25:43 [---] [state_debug] set dirty: Set mode
08-Aug-2008 00:25:43 [---] Processor: 2 AuthenticAMD AMD Athlon(tm) 64 X2 Dual Core Processor 4600+ [x86 Family 15 Model 43 Stepping 1]
08-Aug-2008 00:25:43 [---] Processor features: fpu tsc pae nx sse sse2 3dnow mmx

08-Aug-2008 00:25:43 [---] OS: Microsoft Windows XP: Professional Edition, Servi
ce Pack 2, (05.01.2600.00)
08-Aug-2008 00:25:43 [---] Memory: 2.00 GB physical, 3.85 GB virtual
08-Aug-2008 00:25:43 [---] Disk: 71.56 GB total, 34.54 GB free
08-Aug-2008 00:25:43 [---] Local time is UTC +1 hours
08-Aug-2008 00:25:43 [---] [status_debug] CLIENT_STATE::write_state_file(): Writing state file
08-Aug-2008 00:25:43 [---] [status_debug] CLIENT_STATE::write_state_file(): Done writing state file

E:\Program Files\BOINC>

--- Edit2:

Doesn't look like an environmental problem because the backup copy still works. So presumably something somewhere is corrupted.

--- Edit3:

I deleted all the 'secondary' projects from my client_state.xml, and gave that a go. It worked, but there were a lot of orphaned applications and so forth, so I reverted the client_state.xml and instead I deleted the account files of those other projects. So one of the following projects was causing the trouble: depspid, mindmodelling, gridfinity, harvard clean energy, WCG, and QCN.

---

So one of the following projects was causing the trouble: depspid, mindmodelling, gridfinity, harvard clean energy, WCG, and QCN.

[/quote]

I'm not attached to gridfinity, so that's one down