| 1 |
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN"> |
|---|
| 2 |
<HTML> |
|---|
| 3 |
<HEAD> |
|---|
| 4 |
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=us-ascii"> |
|---|
| 5 |
<META NAME="Generator" CONTENT="MS Exchange Server version 6.5.7036.0"> |
|---|
| 6 |
<TITLE>RE: [mpich2-maint] #403: mpiexec kills the remote login shell</TITLE> |
|---|
| 7 |
</HEAD> |
|---|
| 8 |
<BODY> |
|---|
| 9 |
|
|---|
| 10 |
|
|---|
| 11 |
<P><FONT SIZE=2> Hi,<BR> |
|---|
| 12 |
Can you try out the patch attached. The patch contains some extra debug statements which will help us in narrowing down on your problem.<BR> |
|---|
| 13 |
<BR> |
|---|
| 14 |
Applying the patch<BR> |
|---|
| 15 |
---------------------<BR> |
|---|
| 16 |
1) change directory to top-level of mpich2 source<BR> |
|---|
| 17 |
2) patch -p0 < mpich2_1_0_8_Korebot.patch<BR> |
|---|
| 18 |
3) Re-compile & re-install MPICH2<BR> |
|---|
| 19 |
<BR> |
|---|
| 20 |
Now re-run smpd & mpiexec in debug mode with a simple mpi program, hellow.c (smpd -d > smpd.log / mpiexec -verbose -n 1 ./hellow > mpiexec.log).<BR> |
|---|
| 21 |
<BR> |
|---|
| 22 |
Regards,<BR> |
|---|
| 23 |
Jayesh<BR> |
|---|
| 24 |
<BR> |
|---|
| 25 |
-----Original Message-----<BR> |
|---|
| 26 |
From: Yu-Cheng Chou [<A HREF="mailto:cycchou@ucdavis.edu">mailto:cycchou@ucdavis.edu</A>]<BR> |
|---|
| 27 |
Sent: Thursday, February 12, 2009 1:33 PM<BR> |
|---|
| 28 |
To: Jayesh Krishna<BR> |
|---|
| 29 |
Subject: Re: [mpich2-maint] #403: mpiexec kills the remote login shell<BR> |
|---|
| 30 |
<BR> |
|---|
| 31 |
Hi,<BR> |
|---|
| 32 |
For the first question, I am not able to get the core dump for mpiexec/hellow/ssh on the Korebot because of the limited memory on the Korebot.<BR> |
|---|
| 33 |
<BR> |
|---|
| 34 |
For the second question, I can run such a program with fflush(stdout) and fflush(stderr) statements on the Korebot.<BR> |
|---|
| 35 |
<BR> |
|---|
| 36 |
Thank you<BR> |
|---|
| 37 |
<BR> |
|---|
| 38 |
On Thu, Feb 12, 2009 at 11:14 AM, Jayesh Krishna <jayesh@mcs.anl.gov> wrote:<BR> |
|---|
| 39 |
> Hi,<BR> |
|---|
| 40 |
> I have yet to make the debug module (shouldn't take much time). The<BR> |
|---|
| 41 |
> answers to the questions in my prev email will help me to put in the<BR> |
|---|
| 42 |
> right debug statements.<BR> |
|---|
| 43 |
><BR> |
|---|
| 44 |
> Regards,<BR> |
|---|
| 45 |
> Jayesh<BR> |
|---|
| 46 |
><BR> |
|---|
| 47 |
> -----Original Message-----<BR> |
|---|
| 48 |
> From: Yu-Cheng Chou [<A HREF="mailto:cycchou@ucdavis.edu">mailto:cycchou@ucdavis.edu</A>]<BR> |
|---|
| 49 |
> Sent: Thursday, February 12, 2009 12:23 PM<BR> |
|---|
| 50 |
> To: Jayesh Krishna<BR> |
|---|
| 51 |
> Subject: Re: [mpich2-maint] #403: mpiexec kills the remote login shell<BR> |
|---|
| 52 |
><BR> |
|---|
| 53 |
> Hi,<BR> |
|---|
| 54 |
> Would you give me the debug module directly?<BR> |
|---|
| 55 |
> Thank you<BR> |
|---|
| 56 |
><BR> |
|---|
| 57 |
> On Thu, Feb 12, 2009 at 10:15 AM, Jayesh Krishna <jayesh@mcs.anl.gov> wrote:<BR> |
|---|
| 58 |
>> Hi,<BR> |
|---|
| 59 |
>> Do you get a core dump of mpiexec/hellow/ssh ? (if yes, what does it<BR> |
|---|
| 60 |
>> show<BR> |
|---|
| 61 |
>> ?)<BR> |
|---|
| 62 |
>> Can you run a simple non-MPI C program with fflush(stdout) &<BR> |
|---|
| 63 |
>> fflush(stderr) in it?<BR> |
|---|
| 64 |
>> If the above suggestions don't narrow down the problem I will give<BR> |
|---|
| 65 |
>> you a debug module (patch with some extra printfs) to help us narrow<BR> |
|---|
| 66 |
>> down the problem.<BR> |
|---|
| 67 |
>><BR> |
|---|
| 68 |
>> (PS: I looked into the code, but cannot think of anything that might<BR> |
|---|
| 69 |
>> fail in your environment.)<BR> |
|---|
| 70 |
>><BR> |
|---|
| 71 |
>> Regards,<BR> |
|---|
| 72 |
>> Jayesh<BR> |
|---|
| 73 |
>><BR> |
|---|
| 74 |
>> -----Original Message-----<BR> |
|---|
| 75 |
>> From: Yu-Cheng Chou [<A HREF="mailto:cycchou@ucdavis.edu">mailto:cycchou@ucdavis.edu</A>]<BR> |
|---|
| 76 |
>> Sent: Thursday, February 05, 2009 4:27 PM<BR> |
|---|
| 77 |
>> To: Jayesh Krishna<BR> |
|---|
| 78 |
>> Cc: mpich2-maint@mcs.anl.gov<BR> |
|---|
| 79 |
>> Subject: Re: [mpich2-maint] #403: mpiexec kills the remote login<BR> |
|---|
| 80 |
>> shell<BR> |
|---|
| 81 |
>><BR> |
|---|
| 82 |
>>> Hi,<BR> |
|---|
| 83 |
>>> The debug outputs look normal (the problem could be with the part of<BR> |
|---|
| 84 |
>>> the code at mpiexec exit() which has no dbg statements). I have<BR> |
|---|
| 85 |
>>> added this to our bug tracking list.<BR> |
|---|
| 86 |
>>> Meanwhile,<BR> |
|---|
| 87 |
>>><BR> |
|---|
| 88 |
>>> # Can you send us your ".smpd" config file ?<BR> |
|---|
| 89 |
>><BR> |
|---|
| 90 |
>> The ".smpd" file only contains one line of statement as follows.<BR> |
|---|
| 91 |
>><BR> |
|---|
| 92 |
>> phrase=123<BR> |
|---|
| 93 |
>><BR> |
|---|
| 94 |
>>> # Did you modify the MPICH2 code to run on Korbet (Please send us<BR> |
|---|
| 95 |
>>> your configure command & any env settings set to configure/make<BR> |
|---|
| 96 |
>>> MPICH2)?<BR> |
|---|
| 97 |
>><BR> |
|---|
| 98 |
>> I did not modify the MPICH2 source code.<BR> |
|---|
| 99 |
>> The configure command that I used is listed below.<BR> |
|---|
| 100 |
>><BR> |
|---|
| 101 |
>> ./configure LDFLAGS=-L/tmp/korebot/mpich2-1.0.8/korebot_openssl/lib<BR> |
|---|
| 102 |
>> --host=arm-linux --with-cross=crosstype --with-pm=smpd --with-mpe=no<BR> |
|---|
| 103 |
>> --disable-f90 --disable-f77 --disable-cxx<BR> |
|---|
| 104 |
>> --prefix=/tmp/korebot/mpich2-1.0.8/korebot_mpich2<BR> |
|---|
| 105 |
>><BR> |
|---|
| 106 |
>> The "korebot_openssl/lib" contains the libraries needed for building smpd.<BR> |
|---|
| 107 |
>><BR> |
|---|
| 108 |
>> The content of the file "crosstype" is listed below.<BR> |
|---|
| 109 |
>><BR> |
|---|
| 110 |
>> CROSS_SIZEOF_FLOAT_INT=8<BR> |
|---|
| 111 |
>> CROSS_SIZEOF_DOUBLE_INT=12<BR> |
|---|
| 112 |
>> CROSS_SIZEOF_LONG_INT=8<BR> |
|---|
| 113 |
>> CROSS_SIZEOF_SHORT_INT=8<BR> |
|---|
| 114 |
>> CROSS_SIZEOF_2_INT=8<BR> |
|---|
| 115 |
>> CROSS_SIZEOF_LONG_DOUBLE_INT=16<BR> |
|---|
| 116 |
>><BR> |
|---|
| 117 |
>> Thank you<BR> |
|---|
| 118 |
>><BR> |
|---|
| 119 |
>><BR> |
|---|
| 120 |
>>> > -----Original Message-----<BR> |
|---|
| 121 |
>>> > From: Yu-Cheng Chou [<A HREF="mailto:cycchou@ucdavis.edu">mailto:cycchou@ucdavis.edu</A>] > Sent:<BR> |
|---|
| 122 |
>>> Wednesday, February 04, 2009 2:32 PM > To: Jayesh Krishna > Cc:<BR> |
|---|
| 123 |
>>> mpich-discuss@mcs.anl.gov > Subject: Re: [mpich-discuss] mpiexec<BR> |
|---|
| 124 |
>>> kills the remote login shell > > > Hi,<BR> |
|---|
| 125 |
>>> > > Does smpd abort when you run your MPI job ?<BR> |
|---|
| 126 |
>>> ><BR> |
|---|
| 127 |
>>> > No.<BR> |
|---|
| 128 |
>>> ><BR> |
|---|
| 129 |
>>> > ><BR> |
|---|
| 130 |
>>> > > Regards,<BR> |
|---|
| 131 |
>>> > > Jayesh<BR> |
|---|
| 132 |
>>> > ><BR> |
|---|
| 133 |
>>> > > -----Original Message-----<BR> |
|---|
| 134 |
>>> > > From: Yu-Cheng Chou [<A HREF="mailto:cycchou@ucdavis.edu">mailto:cycchou@ucdavis.edu</A>] > > Sent:<BR> |
|---|
| 135 |
>>> Wednesday, February 04, 2009 1:56 PM > > To: Jayesh Krishna > > Cc:<BR> |
|---|
| 136 |
>>> mpich-discuss@mcs.anl.gov > > Subject: Re: [mpich-discuss] mpiexec<BR> |
|---|
| 137 |
>>> kills the remote login shell > > > > Hi > > > > I can<BR> |
|---|
| 138 |
>>> cross-compile the program and then simply run the > executable on <BR> |
|---|
| 139 |
>>> ><BR> |
|---|
| 140 |
>>> > Korebot with no errors.<BR> |
|---|
| 141 |
>>> > ><BR> |
|---|
| 142 |
>>> > ><BR> |
|---|
| 143 |
>>> > >> Hi,<BR> |
|---|
| 144 |
>>> > >> Can you try running (without mpiexec) a simple C program with<BR> |
|---|
| 145 |
>>> > >><BR> |
|---|
| 146 |
>>> exit(-1) on Korebot ?<BR> |
|---|
| 147 |
>>> > >><BR> |
|---|
| 148 |
>>> > >> ========================================<BR> |
|---|
| 149 |
>>> > >> #include <stdlib.h><BR> |
|---|
| 150 |
>>> > >> int main(int argc, char *argv[]) > >> {<BR> |
|---|
| 151 |
>>> > >> exit(-1);<BR> |
|---|
| 152 |
>>> > >> }<BR> |
|---|
| 153 |
>>> > >> ========================================<BR> |
|---|
| 154 |
>>> > >><BR> |
|---|
| 155 |
>>> > >> Regards,<BR> |
|---|
| 156 |
>>> > >> Jayesh<BR> |
|---|
| 157 |
>>> > >> ________________________________ > >> From:<BR> |
|---|
| 158 |
>>> mpich-discuss-bounces@mcs.anl.gov > >><BR> |
|---|
| 159 |
>>> [<A HREF="mailto:mpich-discuss-bounces@mcs.anl.gov">mailto:mpich-discuss-bounces@mcs.anl.gov</A>] On Behalf Of Jayesh > >><BR> |
|---|
| 160 |
>>> Krishna > >> Sent: Wednesday, February 04, 2009 1:04 PM > >> To:<BR> |
|---|
| 161 |
>>> 'Yu-Cheng Chou'<BR> |
|---|
| 162 |
>>> > >> Cc: mpich-discuss@mcs.anl.gov<BR> |
|---|
| 163 |
>>> > >> Subject: Re: [mpich-discuss] mpiexec kills the remote login<BR> |
|---|
| 164 |
>>> shell ><BR> |
|---|
| 165 |
>>>>> > >> Hi,<BR> |
|---|
| 166 |
>>> > >> Can you also attach the corresponding smpd debug output ?<BR> |
|---|
| 167 |
>>> > >><BR> |
|---|
| 168 |
>>> > >> Regards,<BR> |
|---|
| 169 |
>>> > >> Jayesh<BR> |
|---|
| 170 |
>>> > >><BR> |
|---|
| 171 |
>>> > >> -----Original Message-----<BR> |
|---|
| 172 |
>>> > >> From: Yu-Cheng Chou [<A HREF="mailto:cycchou@ucdavis.edu">mailto:cycchou@ucdavis.edu</A>] > >> Sent:<BR> |
|---|
| 173 |
>>> Wednesday, February 04, 2009 1:02 PM > >> To: Jayesh Krishna > >> Cc:<BR> |
|---|
| 174 |
>>> mpich-discuss@mcs.anl.gov > >> Subject: Re: [mpich-discuss] mpiexec<BR> |
|---|
| 175 |
>>> kills the remote login shell > >> > >> Hi, > >> > >> Firstly,<BR> |
|---|
| 176 |
>>> the previously attached mpiexec verbose output is > a wrong one.<BR> |
|---|
| 177 |
>>> > >> I've attached the correct one to this email.<BR> |
|---|
| 178 |
>>> > >><BR> |
|---|
| 179 |
>>> > >> Secondly, I want to point out that as long as mpiexec is<BR> |
|---|
| 180 |
>>> initiated ><BR> |
|---|
| 181 |
>>>>> from Korebot to run a program, no matter it's a MPI or non-MPI ><BR> |
|---|
| 182 |
>>>>> >><BR> |
|---|
| 183 |
>>> program, no matter the program can be found or not, as soon as > >><BR> |
|---|
| 184 |
>>> mpiexec is finished, the ssh connection to Korebot will be gone.<BR> |
|---|
| 185 |
>>> > >><BR> |
|---|
| 186 |
>>> > >> Thank you<BR> |
|---|
| 187 |
>>> > >><BR> |
|---|
| 188 |
>>> > >><BR> |
|---|
| 189 |
>>> > >>> Hi,<BR> |
|---|
| 190 |
>>> > >>> The mpiexec output shows the following error when<BR> |
|---|
| 191 |
>>> > running hellow,<BR> |
|---|
| 192 |
>>> > >>> ==================<BR> |
|---|
| 193 |
>>> > >>><BR> |
|---|
| 194 |
>>> > >>> Unable to exec 'hello' on korebot > >>> > >>> Error 2 - No<BR> |
|---|
| 195 |
>>> such file or directory > >>> > >>> ================== > >>><BR> |
|---|
| 196 |
>>> > >>> Please provide the debug output of smpd (smpd -d 2>&1 | tee<BR> |
|---|
| 197 |
>>> > >>> smpd.out) along with mpiexec (mpiexec -verbose -n 2 ><BR> |
|---|
| 198 |
>>> ./hellow<BR> |
|---|
| 199 |
>>> 2>&1<BR> |
|---|
| 200 |
>>> | > >>> tee mpiexec.out).<BR> |
|---|
| 201 |
>>> > >>><BR> |
|---|
| 202 |
>>> > >>> # Can you run simple C programs (without using mpiexec) ><BR> |
|---|
| 203 |
>>> on Korbet ?<BR> |
|---|
| 204 |
>>> > >>> # Is the ssh connection aborted when you run non-MPI<BR> |
|---|
| 205 |
>>> programs<BR> |
|---|
| 206 |
>>> > >>> (mpiexec -n 2 > >>> hostname) ?<BR> |
|---|
| 207 |
>>> > >>> # Can you send us your ".smpd" config file ?<BR> |
|---|
| 208 |
>>> > >>> # Did you modify the MPICH2 code to run on Korbet > (Please<BR> |
|---|
| 209 |
>>> send us > >>> your configure command & any env settings set to ><BR> |
|---|
| 210 |
>>> configure/make MPICH2)?<BR> |
|---|
| 211 |
>>> > >>><BR> |
|---|
| 212 |
>>> > >>> Regards,<BR> |
|---|
| 213 |
>>> > >>> Jayesh<BR> |
|---|
| 214 |
>>> > >>><BR> |
|---|
| 215 |
>>> > >>> ________________________________ > >>> From:<BR> |
|---|
| 216 |
>>> mpich-discuss-bounces@mcs.anl.gov > >>><BR> |
|---|
| 217 |
>>> [<A HREF="mailto:mpich-discuss-bounces@mcs.anl.gov">mailto:mpich-discuss-bounces@mcs.anl.gov</A>] On Behalf Of Jayesh ><BR> |
|---|
| 218 |
>>> >>> Krishna > >>> Sent: Wednesday, February 04, 2009 8:41 AM > >>> To:<BR> |
|---|
| 219 |
>>> 'Yu-Cheng Chou'<BR> |
|---|
| 220 |
>>> > >>> Cc: mpich-discuss@mcs.anl.gov > >>> Subject: Re:<BR> |
|---|
| 221 |
>>> [mpich-discuss] mpiexec kills the remote login shell ><BR> |
|---|
| 222 |
>>>>>> > >>> Hi,<BR> |
|---|
| 223 |
>>> > >>> I will take a look at the debug logs and get back to you.<BR> |
|---|
| 224 |
>>> > >>> Meanwhile, can you run simple C programs without using<BR> |
|---|
| 225 |
>>> mpiexec on ><BR> |
|---|
| 226 |
>>>>>> Korbet ?<BR> |
|---|
| 227 |
>>> > >>> MPICH2 currently does not support heterogeneous systems (So you<BR> |
|---|
| 228 |
>>> > >>> won't be able to run your MPI job across ARM & other ><BR> |
|---|
| 229 |
>>> architectures).<BR> |
|---|
| 230 |
>>> > >>><BR> |
|---|
| 231 |
>>> > >>> Regards,<BR> |
|---|
| 232 |
>>> > >>> Jayesh<BR> |
|---|
| 233 |
>>> > >>><BR> |
|---|
| 234 |
>>> > >>> -----Original Message-----<BR> |
|---|
| 235 |
>>> > >>> From: Yu-Cheng Chou [<A HREF="mailto:cycchou@ucdavis.edu">mailto:cycchou@ucdavis.edu</A>] > >>> Sent:<BR> |
|---|
| 236 |
>>> Tuesday, February 03, 2009 7:52 PM > >>> To: Jayesh Krishna > >>> Cc:<BR> |
|---|
| 237 |
>>> mpich-discuss@mcs.anl.gov > >>> Subject: Re: [mpich-discuss]<BR> |
|---|
| 238 |
>>> mpiexec kills the remote login shell > >>> > >>>> # Can you run<BR> |
|---|
| 239 |
>>> non-MPI programs using mpiexec (mpiexec -n > 2 hostname) ?<BR> |
|---|
| 240 |
>>> > >>> Yes.<BR> |
|---|
| 241 |
>>> > >>><BR> |
|---|
| 242 |
>>> > >>>> # Can you compile and run the hello world program > >>>><BR> |
|---|
| 243 |
>>> (examples/hellow.c) provided with MPICH2 (mpiexec -n 2 ./hellow)?<BR> |
|---|
| 244 |
>>> > >>> Yes.<BR> |
|---|
| 245 |
>>> > >>><BR> |
|---|
| 246 |
>>> > >>>> # How did you start smpd (the command used to start > smpd) ?<BR> |
|---|
| 247 |
>>> How did > >>>> you run your MPI job (the command used to run your job)?<BR> |
|---|
| 248 |
>>> > >>> I have a ".smpd" file containing one line of information, ><BR> |
|---|
| 249 |
>>> which is > >>> "phrase=123".<BR> |
|---|
| 250 |
>>> > >>> Thus, I started smpd using "smpd -s".<BR> |
|---|
| 251 |
>>> > >>> Then I used "mpiexec -n 1 hellow" to run hellow on Korebot.<BR> |
|---|
| 252 |
>>> > >>><BR> |
|---|
| 253 |
>>> > >>>> # How did you find that mpiexec kills the sshd process (We <BR> |
|---|
| 254 |
>>> ><BR> |
|---|
| 255 |
>>> >>>> typically ssh to unix machines and run mpiexec without > any<BR> |
|---|
| 256 |
>>> >>>> problems) ?<BR> |
|---|
| 257 |
>>> > >>> I logged in Korebot with two terminals.<BR> |
|---|
| 258 |
>>> > >>> >From #1 terminal, I checked all the processes running on Korebot.<BR> |
|---|
| 259 |
>>> > >>> >From #2 terminal, I started smpd and run hellow using > the<BR> |
|---|
| 260 |
>>> commands > >>> mentioned above.<BR> |
|---|
| 261 |
>>> > >>> After hellow was finished, the connection to Korebot via ><BR> |
|---|
| 262 |
>>> #2 terminal > >>> was closed.<BR> |
|---|
| 263 |
>>> > >>> >From #1 terminal, I knew that the sshd process > associated<BR> |
|---|
| 264 |
>>> with<BR> |
|---|
| 265 |
>>> #2 > >>> >terminal > >>> was gone.<BR> |
|---|
| 266 |
>>> > >>><BR> |
|---|
| 267 |
>>> > >>>> Can you run smpd/mpiexec in debug mode and provide us with<BR> |
|---|
| 268 |
>>> the ><BR> |
|---|
| 269 |
>>>>>>> outputs (smpd -d / mpiexec -n 2 -verbose hostname) ?<BR> |
|---|
| 270 |
>>> > >>> The first attached text file is the output from running<BR> |
|---|
| 271 |
>>> hellow in ><BR> |
|---|
| 272 |
>>>>>> mpiexec's verbose mode.<BR> |
|---|
| 273 |
>>> > >>><BR> |
|---|
| 274 |
>>> > >>><BR> |
|---|
| 275 |
>>> > >>> There is another issue.<BR> |
|---|
| 276 |
>>> > >>> This time, I used two machines. One is Korebot as ><BR> |
|---|
| 277 |
>>> mentioned above, > >>> and the other is a laptop running Ubuntu Linux OS.<BR> |
|---|
| 278 |
>>> > >>> I started smpd with the same ".smpd" file and command as ><BR> |
|---|
| 279 |
>>> mentioned > >>> above both on Korebot and the lap top.<BR> |
|---|
| 280 |
>>> > >>> There is a machine file called "hostfile" on Korebot. The<BR> |
|---|
| 281 |
>>> file<BR> |
|---|
| 282 |
>>> > >>> contains the following information about the name of the ><BR> |
|---|
| 283 |
>>> > >>> two machines.<BR> |
|---|
| 284 |
>>> > >>><BR> |
|---|
| 285 |
>>> > >>> korebot<BR> |
|---|
| 286 |
>>> > >>> shrimp<BR> |
|---|
| 287 |
>>> > >>><BR> |
|---|
| 288 |
>>> > >>> Then from Korebot, I ran cpi using the following command.<BR> |
|---|
| 289 |
>>> > >>><BR> |
|---|
| 290 |
>>> > >>> mpiexec -machinefile ./hostfile -verbose -n 2 cpi > >>> ><BR> |
|---|
| 291 |
>>> >>><BR> |
|---|
| 292 |
>>> ><BR> |
|---|
| 293 |
>>>>>> But the value of pi is a huge number. I think it is related to ><BR> |
|---|
| 294 |
>>>>>> >>><BR> |
|---|
| 295 |
>>> "double type variables" being transferred between > processes<BR> |
|---|
| 296 |
>>> running<BR> |
|---|
| 297 |
>>> ><BR> |
|---|
| 298 |
>>>>>> on an ARM-based Linux and a general Linux machines.<BR> |
|---|
| 299 |
>>> > >>><BR> |
|---|
| 300 |
>>> > >>> The second attached text file is the output from running cpi<BR> |
|---|
| 301 |
>>> in<BR> |
|---|
| 302 |
>>> ><BR> |
|---|
| 303 |
>>>>>> mpiexec's verbose mode.<BR> |
|---|
| 304 |
>>> > >>><BR> |
|---|
| 305 |
>>> > >>><BR> |
|---|
| 306 |
>>> > >>>><BR> |
|---|
| 307 |
>>> > >>>> I am cross-compiling mpich2-1.0.8 with smpd for Khepera ><BR> |
|---|
| 308 |
>>> III mobile > >>>> robot.<BR> |
|---|
| 309 |
>>> > >>>><BR> |
|---|
| 310 |
>>> > >>>> This mobile robot has a Korebot board which is an ARM-based<BR> |
|---|
| 311 |
>>> ><BR> |
|---|
| 312 |
>>> >>>> computer with a Linux operating system.<BR> |
|---|
| 313 |
>>> > >>>><BR> |
|---|
| 314 |
>>> > >>>> The cross-compilation was fine.<BR> |
|---|
| 315 |
>>> > >>>><BR> |
|---|
| 316 |
>>> > >>>> Firstly, I logged in to Korebot through ssh.<BR> |
|---|
| 317 |
>>> > >>>> Secondly, I started smpd.<BR> |
|---|
| 318 |
>>> > >>>> Thirdly, I ran mpiexec to execute an MPI program (cpi) ><BR> |
|---|
| 319 |
>>> that comes > >>>> with the package.<BR> |
|---|
| 320 |
>>> > >>>><BR> |
|---|
| 321 |
>>> > >>>> The result was correct, but when mpiexec was finished, the<BR> |
|---|
| 322 |
>>> ssh<BR> |
|---|
| 323 |
>>> ><BR> |
|---|
| 324 |
>>>>>>> connection to the Korebot was closed.<BR> |
|---|
| 325 |
>>> > >>>> I found that mpiexec kills the sshd process through which I<BR> |
|---|
| 326 |
>>> was ><BR> |
|---|
| 327 |
>>>>>>> remotely connected to Korebot.<BR> |
|---|
| 328 |
>>> > >>>><BR> |
|---|
| 329 |
>>> > >>>> I've been looking for the cause, but still have not > found<BR> |
|---|
| 330 |
>>> any clues.<BR> |
|---|
| 331 |
>>> > >>>><BR> |
|---|
| 332 |
>>> > >>>> Could you give me any ideas to solve this problem?<BR> |
|---|
| 333 |
>>> > >>>><BR> |
|---|
| 334 |
>>> > >>>> Thank you,<BR> |
|---|
| 335 |
>>> > >>>><BR> |
|---|
| 336 |
>>> > >>>> Yu-Cheng<BR> |
|---|
| 337 |
>>> > >>>><BR> |
|---|
| 338 |
>>> > >>><BR> |
|---|
| 339 |
>>> > >><BR> |
|---|
| 340 |
>>> > ><BR> |
|---|
| 341 |
>>> ><BR> |
|---|
| 342 |
>>> }}}<BR> |
|---|
| 343 |
>>><BR> |
|---|
| 344 |
>>><BR> |
|---|
| 345 |
>>> --<BR> |
|---|
| 346 |
>>> Ticket URL: <<A HREF="https://trac.mcs.anl.gov/projects/mpich2/ticket/403">https://trac.mcs.anl.gov/projects/mpich2/ticket/403</A>><BR> |
|---|
| 347 |
>>><BR> |
|---|
| 348 |
>><BR> |
|---|
| 349 |
><BR> |
|---|
| 350 |
</FONT> |
|---|
| 351 |
</P> |
|---|
| 352 |
|
|---|
| 353 |
</BODY> |
|---|
| 354 |
</HTML> |
|---|
