These are chat archives for beniz/deepdetect

28th
Apr 2016
Erkhemee
@erkhemee
Apr 28 2016 00:57
@beniz Ok, I'm opening an issue. Pls, tell me which server output is needed?
my dd is docker version with cpu
Ozéias Sant'ana
@ozeias
Apr 28 2016 04:47
hey @beniz. I'm having the same issue - daemon crashes all the time. One thing that maybe is the problem:
Apr 28 04:44:00 ip-172-30-1-239 kernel: [ 3872.708588] init: deepdetect main process (27839) terminated with status 134 Apr 28 04:44:00 ip-172-30-1-239 kernel: [ 3872.708599] init: deepdetect main process ended, respawning Apr 28 04:44:00 ip-172-30-1-239 nvidia-persistenced: The daemon no longer has permission to remove its runtime data directory /var/run/nvidia-persistenced Apr 28 04:44:00 ip-172-30-1-239 nvidia-persistenced: Shutdown (28109) Apr 28 04:44:02 ip-172-30-1-239 nvidia-persistenced: Started (28556)

this is my memory layer

name: "TestNet" layer { name: "TestNet" type: "MemoryData" top: "data" top: "label" memory_data_param { batch_size: 96 channels: 1 height: 224 width: 224 } }

Emmanuel Benazera
@beniz
Apr 28 2016 07:54
@erkhemee you are correct that there's no access to the server log within the docker container, yet. This feature has been requested before, we'll spawn an issue for this as well
@ozeias the server should never crash, please open an issue, these are things we take very seriously. We'll investigate but I've seen similar nvidia output (without crash), and you should try to do what the kernel tells you, check on permissions over /var/run/nvidia-persistenced. FYI as AFAIK this is an nvidia driver thing, dd by itself does not access any other directory than those where the datasets and models lie.
Ozéias Sant'ana
@ozeias
Apr 28 2016 14:47
@beniz perfect. Here is the issue link beniz/deepdetect#111
thanks and let me know if I can help somehow :)
Emmanuel Benazera
@beniz
Apr 28 2016 14:51
sure, thanks, I'll look at it shortly