Installing BoostIO

Creating a BoostIO Server Running User

Run the following commands on the node where BoostIO is installed to create a user account.

1. Create user group biogroup.

   groupadd -g 1000 biogroup

2. Create user bioadmin in user group biogroup.

   useradd -g 1000 -d /home/bioadmin -u 9000 -m -s /bin/bash bioadmin

(Optional) Cleaning Up the Environment

• Before the installation, ensure that BoostIO is not installed in the current environment. If BoostIO is installed, clear the environment to prepare for the new installation.
• You are advised to delete unused log files from the SDK client in a timely manner to prevent drive space exhaustion.
• The maximum size of a statistics file is 10 MB on the SDK client and 50 MB on the server. Statistics are collected cyclically. After BoostIO is redeployed and started, a new statistics file is generated. You are advised to clear the old statistics file.

1. Collect the IP addresses of the nodes on which you want to install BoostIO.
2. Provide at least one NVMe SSD for each node in the cluster and set the SSD owner to the current installation user and user group.

   chown [Server_installation_user:Server_installation_user_group] /dev/nvmexnx

3. During the initial installation, you need to create the following directories and configure permissions for them:

   Table 1 Directories and permissions

   Directory	User and User Group	Permission	Description
   /opt/boostio	Server_installation_user:Server_installation_user_group	750	BoostIO installation directory.
   /var/log/boostio	Server_installation_user:Server_installation_user_group	750	BoostIO server log directory.
   /var/log/boostio/trace	Server_installation_user:Server_installation_user_group	750	BoostIO statistics log directory.
   /home/ip (This IP address is the same as that in the host_ip_list file on each node.)	Server_installation_user:Server_installation_user_group	750	Directory for storing temporary files during BoostIO installation. The files are automatically deleted after the installation is complete.
   /var/log/jfs	SDK_installation_user:SDK_installation_user_group	750	BoostIO SDK client log directory.

4. Configure the Ceph key ring permission.
   

   Starting BoostIO requires reading the Ceph client key and obtaining the corresponding permission. For details, see the Ceph official website.

5. Run the following command on the ZooKeeper server node (ZooKeeper 3.8.1 as an example) to clear the BoostIO cluster information. Add the SO file required by the ZooKeeper client to the {BoostIO_Home}/lib directory, and change the owner to Server_installation_user:Server_installation_user_group and the permissions to 550.

   sh /install_path/apache-zookeeper-3.8.1-bin/bin/zkCli.sh
   >>deleteall /cm

6. Clear the BoostIO drive management metadata.

   dd bs=8k count=1024 if=/dev/zero of=/dev/nvmexnx

Installing BoostIO

1. Log in to the installation node and upload the ubs_io-boostio-1.0.0-1.{OS_version}.aarch64.rpm software package to any available directory.
2. Install the software package.

   rpm -ivh --nodeps ubs_io-boostio-1.0.0-1.{OS_version}.aarch64.rpm

   After the installation, the following files are generated in the /home directory:

   BoostIO_{version}_Linux-{arch}_release.tar.gz and the boostio directory extracted from the TAR package. For details about the directory structure, see Table 2.

   Table 2 Directory structure of the installation package

   Directory	Folder in the Directory	Description
   BoostIO	bin	Executable file.
   	conf	Configuration file.
   	lib	Binary dependency library.
   	include	Header file.
   	scripts	Tool scripts.

   The tool scripts used to install BoostIO are stored in the scripts directory. See Table 3.

   Table 3 Files in the scripts directory

   Directory	Tool Name	Usage	Execution Method	Parameter
   scripts	hand_out_deploy.py	Installation tool script.	hand_out_deploy.py [option]	install [pkg_path] [user] [group] uninstall
   	host_ip_list	Communication IP address and drive information configuration file.	Used by hand_out_deploy.py.	-
   	install.sh	Installation execution script.	install.sh [option] Used by hand_out_deploy.py.	install [user] [group] [install_path] uninstall
   	scp_file.sh	scp command execution file.	Used by hand_out_deploy.py.	-
   	ssh_cmd.sh	ssh command execution file.	Used by hand_out_deploy.py.	-

   Table 4 Files in the bin directory

   Directory	File Name	Description
   bin	bio_daemon	Executable file of the BoostIO service.
   	seceasy_encrypt	Executable file of the encryption service.

   Table 5 Files in the lib directory

   Directory	File Name	Description
   lib	libbdm.so	Shared object file of BDM, which is used for drive management.
   	libbio_interceptor_server.so	Shared object file of the bridging service.
   	libbio_sdk.so	Share object file of the BoostIO SDK client.
   	libbio_server.so	Shared object file of the BoostIO server.
   	libhcom.so	Shared object file of HCOM, which is used for network transmission.
   	libhcom_static.a	Static library file of HCOM.
   	libhse_cryption.so	Executable file of hseceasy, which is used for encryption.
   	libock_interceptor.so	Shared object file of the bridging service.
   	libock_iofwd_proxy.so	Shared object file of the bridging service.
   	libsecurec.a	Static library file of the encryption service.
   	libexpire_checker.so	Shared object file of SSL certificate check.

3. Configure the installation information.

   Set configuration items of the bio.conf file in the conf directory based on your environment and service requirements. See Table 6.

   Table 6 BoostIO configuration items

   Module	Configuration Item	Description	Default Value	Value/Range	Remarks
   Log	bio.log.level	Log level.	info	debug info warn trace error	-
   Net	bio.net.data.ip_mask	IP address range.	127.0.0.1/24	*.*.*.*/#, where * ranges from 0 to 255 and # ranges from 0 to 32.	When using JuiceFS for big data services, the value of this field must be the same as the IP address corresponding to the host name in the /etc/hosts file.
   	bio.net.data.listen_port	Network communication port on the service plane.	7201	7201 to 7800	-
   	bio.net.data.protocol	Network protocol.	tcp	rdma tcp	-
   	bio.net.rpc.data.busy_polling_mode	Indicates whether to enable busy-polling for Remote Procedure Call (RPC).	false	true false	Available only to RDMA.
   	bio.net.rpc.data.workers_count	Number of worker cores on the RPC data plane.	4	1 to 16	-
   	bio.net.request.executor.thread.num	Number of threads for processing requests at the receive end.	8	8 to 256	-
   	bio.net.request.executor.queue.size	Depth of the request processing queue at the receive end.	1,024	1,024 to 65,535	-
   	bio.net.ipc.data.busy_polling_mode	Indicates whether to enable busy-polling for inter-process communication (IPC).	false	true false	-
   	bio.net.ipc.data.workers_count	Number of worker cores on the IPC data plane.	4	1 to 128	-
   	bio.net.tls.enable.switch	Network security option.	true	true false	Disabling this option may cause information leakage and spoofing risks. If BoostIO is deployed in separated deployment mode, the value of the enableTls parameter transferred by the BoostIO service initialization API must be the same as the value of this configuration item.
   	bio.net.tls.ca.cert.path	Path to the CA certificate.	/path/CA/cacert.pem	The default value is only an example.	If the security option is enabled, the path must be a valid one. If the security function is disabled, the configuration item is not parsed.
   	bio.net.tls.ca.crl.path	Path to the certificate revocation list (CRL) file.	-	-	If the security option is enabled and the certificate needs to be checked whether it has been revoked, the path must be a valid one. If the security function is disabled, the configuration item is not parsed.
   	bio.net.tls.server.cert.path	Path to the certificate file on the server.	/path/server/servercert.pem	The default value is only an example.	If the security option is enabled, the path must be a valid one. If the security function is disabled, the configuration item is not parsed.
   	bio.net.tls.server.key.path	Path to the certificate private key file on the server.	/path/server/serverkey.pem	The default value is only an example.	If the security option is enabled, the path must be a valid one. If the security function is disabled, the configuration item is not parsed.
   	bio.net.tls.server.key.pass.pathPosix	Path to the private key password of the working certificate.	/path/server/server.keypass	The default value is only an example.	If the security option is enabled, the path must be a valid one. If the security function is disabled, the configuration item is not parsed.
   	bio.net.hesc.server.tls.kfs.master.path	Path to the root key generated when encrypting the private key of the working certificate.	/path/server/master/kfsa	The default value is only an example.	If the security option is enabled, the path must be a valid one. If the security function is disabled, the configuration item is not parsed.
   	bio.net.hesc.server.tls.kfs.pass.standby.path	Path to the standby root key generated when encrypting the private key of the working certificate.	/path/server/standby/kfs	The default value is only an example.	If the security option is enabled, the path must be a valid one. If the security function is disabled, the configuration item is not parsed.
   Cache	bio.cache.qos.enable	Flow control option.	true	false true	Enabling this option impairs the extreme performance. You are advised to disable this option in performance test cases.
   	bio.data.crc.enable	Data integrity verification option.	false	false true	Enabling this option increases the data read and write latencies. You are advised to enable this option in fault locating scenarios.
   	bio.segment.size_in_mb	Cache resource granularity.	4	1 to 16	Unit: MB.
   	bio.mem.size_in_gb	Memory capacity used as cache resources.	50	0 to 512	The value cannot exceed the system memory size. Unit: GB. The value 0 indicates that the node does not provide caching.
   	bio.disk.path	List of drives used as cache resources.	/dev/sdxx:/dev/sdyy	-	Separate them using colons (:) if there are multiple drive paths. The current version supports a maximum of four drives.
   	bio.rcache.evict_water_level	Eviction watermark of the read cache.	90	0 to 100	Percentage of the used read cache.
   	bio.cache.mem_read_write_ratio	Read/write resource ratio of the memory.	5:5	0 to 10:10 to 0	-
   	bio.cache.disk_read_write_ratio	Read/write resource ratio of the drives.	5:5	0 to 10:10 to 0	-
   	bio.work.scene	Application scenario flag.	none	none bigdata	Optional. The default value is none. none: There is no usage restriction. bigdata: Used for big data scenarios. Compared with AI scenarios, the main difference is that I/Os are forcibly aligned in big data scenarios.
   	bio.work.io.alignsize	I/O alignment data size.	1	1 to 4,194,304	(Optional) The unit is byte.
   	bio.wcache.evict_water_level	Eviction watermark of the write cache.	0	0 to 100	(Optional) The default value is 0, indicating the percentage of the used write cache.
   	bio.wcache.negotiate.delay	Eviction negotiation delay.	100	50 to 1,000	Optional. The default value is 100 ms. In scenarios that are sensitive to foreground write performance, set this parameter to a larger value to increase the eviction delay. In other scenarios, set this parameter to a smaller value, which accelerates eviction.
   	bio.trace.enable	Process statistics collection option.	true	false true	Enabling this option impairs the extreme performance. You are advised to disable this option in performance test cases.
   Underfs	bio.underfs.file_system_type	Back-end storage system type.	ceph	ceph hdfs	-
   	bio.underfs.ceph.cfg.path	Path to the Ceph configuration file.	/etc/ceph/ceph.conf	This parameter cannot be left empty.	Mandatory when ceph is selected. The value must be an existing path.
   	bio.underfs.ceph.cluster	Ceph cluster name.	ceph	This parameter cannot be left empty.	Mandatory when ceph is selected.
   	bio.underfs.ceph.user	Ceph user.	client.admin	This parameter cannot be left empty.	Mandatory when ceph is selected.
   	bio.underfs.ceph.pool	Ceph data pool.	0:jfspool0,1:jfspool1	This parameter cannot be left empty.	Mandatory when ceph is selected. Use commas (,) to separate multiple parameters.
   	bio.underfs.hdfs.name_node	NameNode of Hadoop.	default:0	*.*.*.*/#, where * ranges from 0 to 255 and # ranges from 0 to 65535.	(Optional) The default value is default:0. The format is IP_address:Port, which indicates the IP address and port specified in the Hadoop configuration file.
   	bio.underfs.hdfs.working_path	Path for storing files in the HDFS system.	/hdfs	It is a valid path that contains 255 or fewer characters.	Optional. The default value is /hdfs.
   CM	bio.cm.initial.nodes_count	Expected number of nodes during cluster initialization.	2	2 to 256	-
   	bio.cm.copy_num	Data redundancy.	2	2	The current software version supports only dual copies.
   	bio.cm.pts_count	Number of partitions.	16	2 to 8,192	-
   	bio.cm.register_timeout_sec	Timeout duration of the ZooKeeper heartbeat check.	20	10 to 60	Unit: s.
   	bio.cm.register_perm_timeout_sec	Time window for determining permanent faults.	60	60 to 600	Unit: s.
   	bio.cm.zk_host	ZooKeeper service node information. Example: 127.0.0.1:2181,127.0.0.2:2181,127.0.0.3:2181 for a ZooKeeper cluster containing three nodes.	-	This parameter cannot be left empty.	The IP address segment used by ZooKeeper must be the same as the service IP address segment.
   Prometheus	bio.prometheus.exposer	IP address and port number of the Prometheus server.	-	*.*.*.*:#, where * ranges from 0 to 255 and # ranges from 0 to 65535.	(Optional)
   	bio.prometheus.scrape_interval_sec	Prometheus sampling frequency.	15	-	(Optional) The unit is second.

4. Configure the host_ip_list file.
   1. Open the host_ip_list file.

      vim boostio/scripts/host_ip_list

   2. Press i to enter the insert mode. Add the following content to the host_ip_list file (replace the variables with the actual ones):

      ip1::BoostIO_communication_IP_address_1::Drive_address_1:Drive_address_2
      ip2::BoostIO_communication_IP_address_2::Drive_address_1:Drive_address_2

   3. Press Esc, type :wq!, and press Enter to save the file and exit.
5. Set the user and user group to which the drives belong.

   chown [Server_installation_user:Server_installation_user_group] Drive address 1
   chown [Server_installation_user:Server_installation_user_group] Drive address 2

6. Run the installation script.

   python3 hand_out_deploy.py install [Path to the installation package that is not decompressed] [Server installation user] [Server installation user group]

   

   • In the openEuler 20.03 OS, you need to install and configure Python 3 to run the installation script.
   • All nodes in the cluster use the same user account and password.
   Enter the server installation user name and password as prompted.

   Figure 1 Command output
   

   After the installation is complete, all BoostIO files and directories are stored in /opt/boostio.