What will be installed:
- xrootd proxy-cache server
- X509 proxy renewer to allow a valid proxy to be always available to read data from AAA
- ❗ port 30443 of the host open toward the clients that will use the cache
- at least 16GB of RAM, but 32 is a recommended minimum
- O(10TB) of disk space for data caching
- 10Gbps connectivity
- docker
- most commonly you will only need the following:
curl -fsSL https://get.docker.com -o get-docker.sh sudo sh get-docker.sh
- docker-compose
- on linux, usually this is done via:
sudo curl -L "https://github.com/docker/compose/releases/download/1.29.2/docker-compose-$(uname -s)-$(uname -m)" -o /usr/local/bin/docker-compose sudo chmod +x /usr/local/bin/docker-compose
- ❗ User or robot x509 certificates into
certs
folder named asusercert.pem
andusercert.key
git clone https://github.com/comp-dev-cms-ita/compose-xrootd cd compose-xrootd mkdir -p certs echo "MYUSER CERTIFICATE" > ./certs/usercert.pem echo "MYUSER CERTIFICATE KEY" > ./certs/usercert.key chmod 600 ./certs/usercert.key
- a valid telegraf token to be able to push metrics to the central InfluxDB
- please contact diego.ciangottinipg.infn.it to obtain one
- then insert it in
telegraf-config/telegraf.conf
where you findtoken = "CHANGEME"
- put your site name in place if the tag
SITENAME HERE
in thetelegraf-config/telegraf.conf
file - allow telegraf to monitor the docker metrics with the following command:
echo "GID=$(stat -c '%g' /var/run/docker.sock)" >> .env
First of all create all the needed directories:
mkdir -p logs
mkdir -p proxy
mkdir -p data
mkdir -p metadata
❗ N.B. if you want to provide an external volume to store data, you can simply mount it as
./data
folder.
In .env
file you can find the main knobs to tweak the cache configuration. Among the others you should adjust the CACHE_RAM_GB
to be between one third and an half of your total machine memory (e.g. for a machine with a total of 4GB of RAM, you could put CACHE_RAM_GB=2
)
Now everything should be ready to go. Bring up the system with:
docker-compose up -d
and monitor the status via a simple docker ps
command.
When everything is in status healthy
(that can take several minutes), you should be able to find the logs of the xrootd daemon on ./logs
folder.
Please follow the WN deployment instructions