From 9e1119005297acd5188608692e98da23be314a93 Mon Sep 17 00:00:00 2001
From: Newnius <newnius.cn@gmail.com>
Date: Thu, 10 Dec 2020 18:50:03 +0800
Subject: [PATCH] update

---
 .DS_Store              | Bin 6148 -> 0 bytes
 docker-gpu.md          | 115 +++++++++++++++++++++++++++++++++++++++++
 sbin/run_agent_mock.sh |  39 ++++++--------
 3 files changed, 132 insertions(+), 22 deletions(-)
 delete mode 100644 .DS_Store
 create mode 100644 docker-gpu.md

diff --git a/.DS_Store b/.DS_Store
deleted file mode 100644
index ffa915f0f35270bc82388e6536ca237787746e46..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 6148
zcmeHK!Ab)$5KY=u(~7tU!5(w*)<f5d^&%{_9=r)FdQfS1X|W6Iu5`B+t(E-^{Ug7?
z-*F~Mi=_&lMC=Sq-efY9khe=FVT|$Su<I~pGsXlcVxa`h7lPxc3sNv1M2>R=qx3F_
zk^uT|3(@5GiwxkmvslUk7P9#B_Y3Z#FwJVUcU~x3E30clltp=?eC_p<k(c?|Fm3vS
zOYEKcSwArLS&~FE+Yipe-l%PF9VKz*hjDM9lEZEfLar{txSNcc$uRDwY93Dy2vHKH
zwp|&I>&`({)_0vrRgN2tT2=1V_a~E**xqptPdc{`kK?Dw^NVJc!w;xr)8G_d!I*9E
z-W$YG65oJt8b6IABnF59Vql3FFh`(RU!pzJR*3;(;Aafr`5-|NU4ywsb#y?3*GC+;
z5K+L!w*;az=o-v5!Uzc0sen3_n<oa>>EM?p&NY~8)ai_?m0=#UGB+<2u2u)XRN;)f
z8mT1)h=F+qin?oK{XhQx{Xd^XJz{_u_*V>YtL3$tuq9Jl7dD5rR)TgwQ7|soI7<OT
hm12m+Qd|L50)B}GpldMK2p$l+2q+q;AqIYxfe&D(P^ka_

diff --git a/docker-gpu.md b/docker-gpu.md
new file mode 100644
index 0000000..ae3dcf1
--- /dev/null
+++ b/docker-gpu.md
@@ -0,0 +1,115 @@
+https://www.nvidia.com/Download/index.aspx
+
+https://github.com/NVIDIA/nvidia-docker/wiki/Installation-(version-2.0)
+
+
+
+https://github.com/NVIDIA/nvidia-docker/wiki/GPU-isolation-(version-1.0)
+
+
+
+https://unix.stackexchange.com/questions/249643/gpu-usage-per-process-on-a-linux-machine-cuda
+
+
+
+
+
+
+
+sudo yum install -y wget gcc pciutils git
+
+sudo yum install -y kernel-devel-$(uname -r) kernel-headers-$(uname -r)
+
+
+wget http://us.download.nvidia.com/tesla/410.129/NVIDIA-Linux-x86_64-410.129-diagnostic.run
+
+sudo bash NVIDIA-Linux-x86_64-410.129-diagnostic.run
+
+
+curl -fsSL https://get.docker.com | sh
+
+sudo usermod -aG docker $USER
+
+
+distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
+curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.repo | sudo tee /etc/yum.repos.d/nvidia-docker.repo
+
+sudo yum install -y nvidia-container-toolkit
+sudo systemctl restart docker
+
+
+
+git clone https://github.com/tensorflow/models.git
+
+docker run --gpus all -it --rm --mount type=bind,src=/home/newnius_cn/models,dst=/data tensorflow/tensorflow:1.14.0-gpu bash
+
+docker run --gpus '"device=0,1"' -it --rm --mount type=bind,src=/home/newnius_cn/models,dst=/data tensorflow/tensorflow:1.14.0-gpu bash
+
+export PYTHONPATH="$PYTHONPATH:/data"
+
+pip install requests
+
+# mnist
+python official/r1/mnist/mnist_test.py --benchmarks=.
+
+# CIFAR-10
+
+python official/r1/resnet/cifar10_download_and_extract.py --data_dir /data/ds
+
+python official/r1/resnet/cifar10_main.py --data_dir /data/ds/cifar-10-batches-bin --model_dir /tmp
+
+
+
+
+
+```python
+config = tf.ConfigProto()
+config.gpu_options.allow_growth = True
+config.log_device_placement = True
+
+sess = tf.Session(config=config)
+
+tf.keras.backend.set_session(sess)
+```
+
+
+
+
+sudo yum erase -y nvidia-container-runtime-hook.x86_64 nvidia-container-runtime.x86_64 nvidia-docker2.noarch docker-ce docker-ce-cli.x86_64
+
+sudo gpg --homedir /var/lib/yum/repos/x86_64/7/libnvidia-container/gpgdir --delete-key F796ECB0
+
+sudo gpg --homedir /var/lib/yum/repos/x86_64/7/nvidia-container-runtime/gpgdir --delete-key F796ECB0
+
+sudo gpg --homedir /var/lib/yum/repos/x86_64/7/nvidia-docker/gpgdir --delete-key F796ECB0
+
+[Can not use nvidia-container-runtime repository - repomod.xml signature could not be verified](https://github.com/NVIDIA/nvidia-docker/issues/836)
+
+
+
+ERROR: Unable to load the kernel module 'nvidia.ko'. This happens most frequently when this kernel module was built against the wrong or improperly configured kernel sources, with a version of gcc that differs from the one used to build the target kernel, or if a driver such as rivafb, nvidiafb, or nouveau is present and prevents the NVIDIA kernel module from obtaining ownership of the NVIDIA graphics device(s), or no NVIDIA GPU installed in this system is supported by this NVIDIA Linux graphics driver release.
+
+
+The problem is likely due to one of the issues listed in that error message. You'll need to go through them one by one, modifying things to rule them out.
+
+blacklisting nouveau is frequently not enough. Often it is necessary to remove it from the initrd image as well.
+
+as root:
+
+echo -e "blacklist nouveau\noptions nouveau modeset=0" > /etc/modprobe.d/disable-nouveau.conf
+
+update-initramfs -u
+
+如果提示command not found，似乎没有关系，重启机器即可。
+
+
+It looks like you did something like the first line already - that is the blacklist. The second line will remove it from the initrd image.
+
+Section 8.1 of the driver README document discusses nouveau:
+
+http://us.download.nvidia.com/XFree86/Linux-x86_64/340.24/README/index.html
+
+
+
+
+
diff --git a/sbin/run_agent_mock.sh b/sbin/run_agent_mock.sh
index 53ae90d..7b21f7c 100644
--- a/sbin/run_agent_mock.sh
+++ b/sbin/run_agent_mock.sh
@@ -1,26 +1,21 @@
 #!/bin/bash
 
+# Example: NUM=100 bash sbin/run_agent_mock.sh
+
 # docker ps -a | grep yao-agent:mock | awk '{print $1}' | xargs docker rm -f
 
-for ((i=1; i<=${NUM}; i++));
-do
-	docker run \
-		--name yao-agent-mock-$i \
-		--network yao-net \
-		--network-alias node-mock-$i \
-		--hostname node-mock-$i \
-		-d \
-		--restart always \
-		--detach=true \
-		--env NUMS=100 \
-		--env ClientHost=$(hostname) \
-		--env HeartbeatInterval=5 \
-		--env ReportAddress='http://yao-scheduler:8080/?action=agent_report' \
-		--env PYTHONUNBUFFERED=1 \
-		--mount type=bind,source=/etc/localtime,target=/etc/localtime,readonly \
-		--mount type=bind,src=/var/run/docker.sock,dst=/var/run/docker.sock \
-		quickdeploy/yao-agent:mock
-
-		# reduce instant load
-		sleep 1
-done
+docker run \
+	--name yao-agent-mock-$(hostname) \
+	--network yao-net \
+	--network-alias node-mock-$(hostname) \
+	--hostname node-mock-$(hostname) \
+	-d \
+	--restart always \
+	--detach=true \
+	--env NUMS=${NUM} \
+	--env ClientHost=node-mock-$(hostname) \
+	--env ReportAddress='http://yao-scheduler:8080/?action=agent_report' \
+	--env PYTHONUNBUFFERED=1 \
+	--mount type=bind,source=/etc/localtime,target=/etc/localtime,readonly \
+	--mount type=bind,src=/var/run/docker.sock,dst=/var/run/docker.sock \
+	quickdeploy/yao-agent:mock