adios_read_open_file seems to stall when not finding file

[fouriaux]

adios_read_open_file seems to stall when not finding file

I try to do a restart feature in my program: I want to check if a file success to open for reading variables and continue if not. However my program seems to block completely after adios_read_open_file fail (which should return NULL and continue).

I run my execution via this command line:

srun -n4 frozen_read 

If I run my program without srun, it is behaving properly: the adios_read_open_file call return NULL as expected. Please find below my code...

frozen_read.cc.txt
config.xml.txt
CMakeLists.txt

[pnorbert]

Uhh, I may have introduced a new bug. I will test this when I am back from the holiday weekend.

…

On Mon, May 29, 2017 at 11:39 AM Jeremy FOURIAUX ***@***.***> wrote: adios_read_open_file seems to stall when not finding file I try to do a restart feature in my program: I want to check if a file success to open for reading variables and continue if not. However my program seems to block completely after adios_read_open_file fail (which should return NULL and continue). I run my execution via this command line: srun -n4 frozen_read If I run my program without srun, it is behaving properly: the adios_read_open_file call return NULL as expected. CMakeLists.txt <https://github.com/ornladios/ADIOS/files/1036179/CMakeLists.txt> — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#131>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADGMLcI9lUYrMDt9Bkypq6XQoFnalo92ks5r-ua9gaJpZM4NpcNG> .

[fouriaux]

ok :-)
I forgot to mention about the error message that is returned:

ERROR: MPI open failed for not_existing.bp: 'File does not exist, error stack:
ADIOI_UFS_OPEN(69): File not_existing.bp does not exist'
ERROR: File open failed: not_existing.bp

[fouriaux]

This proble does appears only with adios_read_open_file: adios_read_open function have correct behavior.

[pnorbert]

Fixed this bug. Can you give it a try? Thanks.

[fouriaux]

Thanks ! It is working 👍
However I got some remaining error messages when I disable Verbosity of the method:

srun -n4 ./1_frozen_read
Invalid error code (269009701) (error ring index 1 invalid)
Invalid error code (269009701) (error ring index 1 invalid)
Invalid error code (269009701) (error ring index 1 invalid)
Invalid error code (269009701) (error ring index 1 invalid)
Invalid error code (269009701) (error ring index 1 invalid)
Invalid error code (269009701) (error ring index 1 invalid)

I am not sure if it is related..

[pnorbert]

I cannot reproduce this message. You meant setting "verbose=0" in adios_read_init_method(), right? My run is quiet in that case.

[fouriaux]

yes for setting verbose=0.
Ok, so it is probably something with my configuration, I will create a new ticket if I don't find something on that.
Thanks a lot.