| RE: [ptp-dev] sdm/gdb Failed to get stack frames from backend |
|
I can downgrade gdb. I’ll let you know if
that fixes it. Brian Henerey From:
ptp-dev-bounces@xxxxxxxxxxx [mailto:ptp-dev-bounces@xxxxxxxxxxx] On Behalf Of Greg Watson Looks like there's a compatibility issue with gdb-6.5 (PTP 1.0 was
built prior to the 6.5 release). I'll see if I can fix and do a 1.0.1
build. Greg On Sep 15, 2006, at 2:33 PM, Henerey, Brian wrote:
Your Control System Choice: 'Open Runtime Environment (ORTE)' Your Monitoring System Choice: 'Open Runtime Environment (ORTE)' refreshRuntimeSystems SHUTTING DOWN CONTROL/MONITORING/PROXY systems where appropriate OMPIProxyRuntimeClient - firing up proxy, waiting for
connecting. Please wait! This can take a minute . . . ORTE_SERVER path =
'/home/bch/Desktop/eclipse/plugins/org.eclipse.ptp.orte.linux.x86_64_1.0.0/bin/orte_server' sessionCreate(0) bind(0.0.0.0/0.0.0.0:0) port=43589 accept thread starting... Waiting on accept. OMPIProxyRuntimeClient waiting on {201, 210} RUNNING PROXY SERVER COMMAND:
'/home/bch/Desktop/eclipse/plugins/org.eclipse.ptp.orte.linux.x86_64_1.0.0/bin/orte_server
--port=43589' AbstractProxyRuntimeClient.handleEvent() got event EVENT_CONNECTED OMPIProxyRuntimeClient got event: EVENT_RUNTIME_CONNECTED OMPIProxyRuntimeClient notifying... OMPIProxyRuntimeClient awoke! <0000000b STARTDAEMON> OMPIProxyRuntimeClient waiting on {200, 201} accept thread exiting... event thread starting... ++++++++++ orte_server: proxy_svr_connect returned. ++++++++++ orte_server: StartDaemon(orted orted --scope public
--seed --persistent --universe PTP-ORTE-32375) ++++++++++ orte_server: Starting execv now! ++++++++++ orte_server: proxy_svr_connect returned. ++++++++++ orte_server: PARENT: orted_pid = 32375 ++++++++++ orte_server: ORTEInit (PTP-ORTE-32375) AbstractProxyRuntimeClient.handleEvent() got event EVENT_OK 200 OMPIProxyRuntimeClient got event: EVENT_RUNTIME_OK OMPIProxyRuntimeClient notifying... OMPIProxyRuntimeClient awoke! OMPIMonitoringSystem startup() JAVA OMPI: getMachines() called OMPIMonitoringSystem: getNodes(machine0) called <0000000f GETNODES 3:3000> OMPIProxyRuntimeClient waiting on {201, 207} AbstractProxyRuntimeClient.handleEvent() got event EVENT_OK 207 0 OMPIProxyRuntimeClient got event: EVENT_RUNTIME_NODES 0 OMPIProxyRuntimeClient notifying... OMPIProxyRuntimeClient awoke! MACHINE: machine0 - #nodes = 1 <000000da GETNATTR 3:3000 5:2d3100
33:4154545249425f4e4f44455f4e414d4500 33:4154545249425f4e4f44455f5553455200
35:4154545249425f4e4f44455f47524f555000 35:415454524942 5f4e4f44455f535441544500 33:4154545249425f4e4f44455f4d4f444500> OMPIProxyRuntimeClient waiting on {201, 208} ++++++++++ orte_server: str = 'OMPI_MCA_universe=PTP-ORTE-32375' ++++++++++ orte_server: Start daemon returning OK. ++++++++++ orte_server: BEFORE CALL KEYS[0] = 'orte-node-name' ++++++++++ orte_server: BEFORE CALL KEYS[1] = 'orte-node-bproc-user' ++++++++++ orte_server: BEFORE CALL KEYS[2] =
'orte-node-bproc-group' ++++++++++ orte_server: BEFORE CALL KEYS[3] =
'orte-node-bproc-status' ++++++++++ orte_server: BEFORE CALL KEYS[4] = 'orte-node-bproc-mode' ++++++++++ orte_server: MAX = 0, MIN = 0 ++++++++++ orte_server: totlen = 0 ++++++++++ orte_server: totlen = 0 ++++++++++ orte_server: Hostname = 'wuosglinux' ++++++++++ orte_server: Username = 'bch' ++++++++++ orte_server: Groupname = 'users' ++++++++++ orte_server: valSTR = 'B:77756F73676C696E757800
4:62636800 6:757365727300' AbstractProxyRuntimeClient.handleEvent() got event EVENT_OK 208
B:77756F73676C696E757800 4:62636800 6:757365727300 OMPIProxyRuntimeClient got event: EVENT_RUNTIME_PROCATTR
[Ljava.lang.String;@d1a9f20 OMPIProxyRuntimeClient notifying... OMPIProxyRuntimeClient awoke! *** attribs[0] = wuosglinux *** attribs[1] = bch *** attribs[2] = users NodeName According to ORTE = 'wuosglinux' #attribs returned: 3 ------------ job monitoringSystemChangeEvent OS = 'linux', Architecture = 'x86_64', OS_ARCH combo =
'linux.x86_64' PTP Version = 1.0.0 Fragment 0 path:
'/home/bch/Desktop/eclipse/plugins/org.eclipse.ptp.debug.sdm.orte.linux.x86_64_1.0.0/' Correct fragment for our
OS & arch Searching for file in
'/home/bch/Desktop/eclipse/plugins/org.eclipse.ptp.debug.sdm.orte.linux.x86_64_1.0.0/bin/sdm'
**** FOUND IT! sessionCreate(0) bind(0.0.0.0/0.0.0.0:0) port=36043 JAVA OMPI: run() with args: name:
Debug/newParallelHelloTest path:
/home/bch/workspace.new/newParallelHelloTest cwd:
/home/bch/workspace.new/newParallelHelloTest machineName: machine0 #procs: 1 #proc/node: 1 firstNode#: 0 isDebug?
true <000003a3 RUN 17:657865634e616d6500
53:44656275672f6e6577506172616c6c656c48656c6c6f5465737400
21:70617468546f4578656300 89:2f686f6d652f6263682f776f726b73706163652e6e6 5772f6e6577506172616c6c656c48656c6c6f5465737400
21:6e756d4f6650726f637300 3:3100 25:70726f63735065724e6f646500 3:3100
25:66697273744e6f64654e756d00 3:3000 21:776f726b 696e6744697200
89:2f686f6d652f6263682f776f726b73706163652e6e65772f6e6577506172616c6c656c48656c6c6f5465737400
25:64656275676765725061746800 183:2f686f6d652f6263682f446 5736b746f702f65636c697073652f706c7567696e732f6f72672e65636c697073652e7074702e64656275672e73646d2e6f7274652e6c696e75782e7838365f36345f312e302e302f62696e2f73646d00
23:6 46562756767657241726700 33:2d2d686f73743d6c6f63616c686f737400
23:646562756767657241726700 35:2d2d64656275676765723d6764622d6d6900
23:646562756767657241726700 57:2d2d6 4656275676765725f706174683d2f7573722f62696e2f67646200
23:646562756767657241726700 25:2d2d706f72743d333630343300> OMPIProxyRuntimeClient waiting on {201, 211} accept thread starting... ++++++++++ orte_server: DeBUG! ++++++++++ orte_server: (debug ? 1) Spawning 1 processes of job
'/home/bch/workspace.new/newParallelHelloTest/Debug/newParallelHelloTest' ++++++++++
orte_server: program name
'Debug/newParallelHelloTest' AbstractProxyRuntimeClient.handleEvent() got event EVENT_OK 211 2 OMPIProxyRuntimeClient got event: EVENT_RUNTIME_NEWJOB jobid=2 OMPIProxyRuntimeClient notifying... OMPIProxyRuntimeClient awoke! ModelManager.run() - new JobID = 2 MODEL MANAGER: newJob(2) getProcsStatusForNewJob:job2 - #procs = 1 <00000078 GETPATTR 3:3200 5:2d3100
37:4154545249425f50524f434553535f50494400
49:4154545249425f50524f434553535f4e4f44455f4e414d4500> OMPIProxyRuntimeClient waiting on {201, 205} ++++++++++ orte_server: SPAWNED [error code 0 = 'Success'], now
unlocking ++++++++++ orte_server: NEW JOBID = 2 ++++++++++ orte_server: res = '211 2' ++++++++++ orte_server: Returning from ORTERun ++++++++++ orte_server: ORTEGetProcessAttribute! ++++++++++
orte_server: jobid = 2, procid
= -1 ++++++++++ orte_server: BEFORE CALL KEYS[0] = 'orte-proc-pid' ++++++++++ orte_server: BEFORE CALL KEYS[1] = 'orte-node-name' ++++++++++ orte_server: AFTER CALL! VALS[0] = '0' ++++++++++ orte_server: AFTER CALL! VALS[1] = 'localhost' AbstractProxyRuntimeClient.handleEvent() got event EVENT_OK 205
2:3000 A:6C6F63616C686F737400 OMPIProxyRuntimeClient got event: EVENT_RUNTIME_PROCATTR
[Ljava.lang.String;@6a844689 OMPIProxyRuntimeClient notifying... OMPIProxyRuntimeClient awoke! *** attribs[0] = 0 *** attribs[1] = localhost ------------ job run: job2 ***** CURRENT COMMAND:
org.eclipse.ptp.debug.external.core.commands.StartDebuggerCommand@5c64a871 AbstractProxyDebugClientgot event EVENT_CONNECTED <000000f9 INI
53:44656275672f6e6577506172616c6c656c48656c6c6f5465737400
89:2f686f6d652f6263682f776f726b73706163652e6e65772f6e6577506172616c6c656c48656c6c6f5465737400
89:2f686f6d652f6263682f776f726b73706163652e6e65772f6e6577506172616c6c656c48656c6c6f5465737400> accept thread exiting... event thread starting... AbstractProxyDebugClientgot event EVENT_OK 111 1:1 got debug event: EVENT_DBG_OK 01 ***** CURRENT COMMAND:
org.eclipse.ptp.debug.external.core.commands.SetFunctionBreakpointCommand@780eb73e <0000004a SFB 9:313a303100 3:3000 3:3100 3:3000 1:00
9:6d61696e00 1:00 3:3000 3:3000> AbstractProxyDebugClientgot event EVENT_OK 104 1:1 0 1 0 0 0
B:627265616B706F696E7400 10:2E2E2F68656C6C6F776F726C642E6300 5:6D61696E00
13:3078303030303030303030303430
3039363700 8 0 got debug event: EVENT_DBG_BPSET 01 ***** CURRENT COMMAND:
org.eclipse.ptp.debug.external.core.commands.GoCommand@38e02fc4 --- Abs debugger:
org.eclipse.ptp.debug.external.core.cdi.event.InferiorResumedEvent@90dae16 <00000010 GOP 9:313a303100> ===================== event:
org.eclipse.ptp.debug.external.core.cdi.event.InferiorResumedEvent@90dae16 --- TESTING err event and remove annotation of processes in this
event: 1 AbstractProxyDebugClientgot event EVENT_OK 112 1:1 4
38:4465627567676572206572726F723A204661696C656420746F2067657420737461636B206672616D65732066726F6D206261636B656E64
00 got debug event: EVENT_DBG_ERROR 01 4 Debugger error: Failed to get
stack frames from backend ======================= EVENT_DBG_ERROR ==================== From: ptp-dev-bounces@xxxxxxxxxxx [mailto:ptp-dev-bounces@xxxxxxxxxxx]
On Behalf Of Greg Watson Can you attach the
rest of the messages from your xterm? Thanks, Greg On Sep 15, 2006,
at 1:47 PM, Henerey, Brian wrote:
mpicc -I/usr/local/include/openmpi/ompi/mpi/cxx
-I/usr/local/include/openmpi/ompi -O0 -g -c -fmessage-length=0 It was at debug level 3, -g3, by default, but I just lowered it to
–g and still have the problem. I also lowered the processes to 1, and still get the same error.
SDM now uses 100% of 1 CPU until I killall sdm. Thanks for your help, Brian From: ptp-dev-bounces@xxxxxxxxxxx
[mailto:ptp-dev-bounces@xxxxxxxxxxx]
On Behalf Of Greg Watson Hi Brian, Did you compile
the program with the -g option? I've seen this error when there are no
debugging symbols in the executable. If you try and
debug a 10 process job on your local machine then you will see 10 copies of the
SDM. The SDM is designed for debugging on real clusters where you would
normally only have one SDM process per machine (or node). There is nothing
wrong with debugging on a local machine, but is probably wise to keep the
number of processes low. There were some issues in the 1.0 release of PTP that
would result in the SDM pegging the CPU. If this happens, just kill off the SDM
and orte_server processes and restart your Eclipse session. Regards, Greg On Sep 15, 2006,
at 1:21 PM, Henerey, Brian wrote:
I’m having a
problem with sdm/gdb when I try to debug a parallel hello world file. I launch
eclipse –clean from an xterm so I can watch the output. I see this error
message: got debug
event: EVENT_DBG_ERROR 03ff 4 Debugger error: Failed to get stack frames from
backend =======================
EVENT_DBG_ERROR ==================== I hope
that’s enough output, otherwise I can post the complete output. My program will
run fine in eclipse, but the debug does not. I have verified that the program
runs from the command line fun with both kdbg and ddd by issuing the following
commands. #mpirun -np
2 -debug ddd newParallelHelloTest #mpirun -np
2 -debug kdbg newParallelHelloTest when I do: #mpirun -np
2 -debug gdb newParallelHelloTest it opens a
(gdb) prompt. Typing run does not seem to do anything. However, if
I run #gdb
newParallelHelloTest And execute
‘(gdb) run’, it will correctly run a single job saying ‘hello, I am 0 of 1’ I’m not
familiar with gdb (or any of these debuggers), but I was trying to establish if
gdb would run my code okay outside of eclipse. I have
verified this problem with the same ‘Failed to get stack frames from backend’
error on 3 machines. 2 run 64 bit Gentoo Linux, the other runs 32 bit Gentoo. The main
system I’ve tested on is my own, which runs 64 bit Gentoo Linux Dual
Opteron, 4GB’s of RAM Linux
version 2.6.17-gentoo-r4 (root@wuosglinux) (gcc version 3.4.6 (Gentoo 3.4.6-r1,
ssp-3.4.5-1.0, pie-8.7.9)) #8 SMP Gdb version
6.5 I download
eclipse 3.2 from the website and run in out of my home directory. I installed
CDT through the eclipse update manager. I installed these 4 additional files: org.eclipse.ptp-1.0.0.tar.gz
org.eclipse.ptp.debug-orte-linux-x86_64-1.0.0.tar.gz
org.eclipse.ptp.mpi-1.0.tar.gz
org.eclipse.ptp.orte-linux-x86_64-1.0.0.tar.gz I was able
to install to of them with the eclipse update manager. The other two untar into
the eclipse/ path. The eclipse
ptp debug perspective does not allow me to step in or step through a running
debug job. The options are grayed out. If I try to launch 10 processes of my
test hello world, then 10 sdm’s start running and use 100% of both my cpus. Does anyone
know what’s causing this or have a suggestion what to try? Thanks, Brian Brian Henerey _______________________________________________ ptp-dev mailing
list _______________________________________________ ptp-dev mailing
list _______________________________________________ ptp-dev mailing list |