Host standardization checks|V4.3.6|OceanBase Cloud Platform|OCP docs|Distributed Database

Host standardization checks

Last Updated：2026-08-05 09:41:04 Updated

During the host standardization process, the system will perform checks on the host. This topic describes the checks performed during the host standardization process.

Note

Host standardization checks are applicable only to hosts where OBServer is deployed. For other host scenarios, please use them as needed.

In a container environment, due to limitations such as cgroup and namespace, checks or automatic fixes may not be possible. We recommend that you refer to the document content and perform the checks as needed.

Check the file system type

The file system type supports EXT4 and XFS. When the disk capacity exceeds 16 TB, check whether the file system type is XFS. If not, we recommend that you reconfigure the disk.

Check the kernel parameters

Check whether the kernel parameters meet the requirements:

Kernel parameter	Default value range	Description
net.core.somaxconn	2048 ≤ value ≤ 16384	The maximum length of the listening queue for each port. This is a global parameter.
net.core.netdev_max_backlog	500 ≤ value ≤ 10000	The maximum number of data packets that can be sent to the queue when the rate at which data packets are received by each network interface exceeds the rate at which the kernel processes these packets.
net.core.rmem_default	65536 ≤ value ≤ 16777216	The default size of the TCP receive window (in bytes).
net.core.wmem_default	65536 ≤ value ≤ 16777216	The default size of the TCP send window (in bytes).
net.core.rmem_max	8388608 ≤ value ≤ 16777216	The maximum size of the TCP receive window (in bytes).
net.core.wmem_max	8388608 ≤ value ≤ 16777216	The maximum size of the TCP send window (in bytes).
net.ipv4.conf.default.rp_filter	== 1	Defines the rules for reverse route verification performed by the network card on received data packets. 0: Reverse route verification is disabled.; 1: Strict reverse route verification is enabled.; 2: Loose reverse route verification is enabled.
net.ipv4.conf.default.accept_source_route	==0	`accept_source_route` specifies whether the sender of a data packet can specify the path for sending the data packet and the path for returning the data packet to the sender. br0 disables this feature.
net.ipv4.tcp_syncookies	==1	When the SYN queue overflows, send syncookies to the other party to prevent SYN flood attacks.
net.ipv4.tcp_rmem	The three values 4096, 87380, and 16777216 are `min`, `default`, and `max`, respectively, and are divided into three parts: 4096 ≤ min ≤ 8192 65536 ≤ default ≤ 174760 8388608 ≤ max ≤ 16777216	Defines the memory used by sockets for automatic tuning. The first value is the minimum number of bytes allocated to the socket receive buffer; The second value is the default value (which will be overridden by rmem_default), and the buffer can grow to this value when the system load is not high; The third value is the maximum number of bytes for the receive buffer (which will be overridden by rmem_max).
net.ipv4.tcp_wmem	The three values 4096, 65536, and 16777216 are divided into three parts: 4096 ≤ min ≤ 8192 65536 ≤ default ≤ 131072 8388608 ≤ max ≤ 16777216	Defines the memory used by sockets for automatic tuning. The first value is the minimum number of bytes allocated to the socket send buffer; The second value is the default value (which will be overridden by wmem_default), and the buffer can grow to this value when the system load is not high; The third value is the maximum number of bytes for the send buffer (which will be overridden by wmem_max).
net.ipv4.tcp_max_syn_backlog	1024 ≤ value ≤ 16384	The maximum number of connection requests that can be saved in the queue before they receive confirmation from the other party. If the server frequently overloads, we recommend that you increase this value. The default value is 1024.
net.ipv4.tcp_fin_timeout	15 ≤ value ≤ 60	The time (in seconds) that a socket connection remains in the FIN-WAIT-2 state after the local end disconnects. The other party may disconnect the connection, or the connection may remain open indefinitely, or the process may unexpectedly terminate.
net.ipv4.tcp_tw_reuse	==1	Specifies whether to allow sockets in the TIME-WAIT state (with TIME-WAIT ports) to be used for new TCP connections.
net.ipv4.tcp_slow_start_after_idle	==0	Specifies whether to enable slow start after TCP is idle. 0 indicates that slow start is disabled.
net.ipv4.ip_local_port_range	3500 ≤ value ≤ 65535	This parameter specifies the range of local port numbers that can be used. We recommend that you set the value to 3500~65535.
net.ipv4.ip_local_reserved_ports	62881 ≤ value ≤ 63881	This parameter specifies the port numbers that are reserved and will not be automatically assigned by the system. Check whether ports in the 62881~63881 range are reserved to ensure that the ocp-agent can use these port numbers to provide O&M and monitoring services.
vm.swappiness	==0	Specifies the relative weight of memory pages to be swapped out.
vm.min_free_kbytes	32768 ≤ value ≤ 2097152	Specifies the size of the reserved idle page pool. We recommend that you set the value to 2097152.
vm.max_map_count	327680 ≤ value ≤ 1000000	Specifies the maximum number of memory-mapped areas that a process can use. The default value is 65536. If your application requires more memory-mapped areas, we recommend that you increase this value to 655360.
vm.overcommit_memory	==0	Specifies the condition for accepting or rejecting large memory requests. The default value is 0.
vm.nr_hugepages	==0	If the value of the `nr_hugepages` parameter is 0, HugePages is not enabled.
fs.aio-max-nr	≥ 1048576	The maximum number of asynchronous I/O operations that can be supported.
kernel.numa_balancing	==0	Disables automatic NUMA balancing to avoid performance jitter caused by memory node data migration.
vm.zone_reclaim_mode	==0	Disables the memory zone reclaim mode. This way, memory can be reclaimed from other zones or NUMA nodes, avoiding performance jitter caused by memory reclaim.
fs.file-max	≥ 6573688	The maximum number of files that can be opened in the entire system. This is a system-level limit. We recommend that you set the value to a value greater than the default value to avoid restrictions.
fs.pipe-user-pages-soft	== 0	The soft limit on the total size (in pages) of all pipes created or set by a non-privileged user (a user without the `CAP_SYS_RESOURCE` or `CAP_SYS_ADMIN` capability). We recommend that you set this parameter to 0 to disable the soft limit.

Manual modification method: Take the net.core.somaxconn parameter as an example. You can write net.core.somaxconn = 128 to the custom configuration file /etc/sysctl.d/oceanbase.conf by using the echo command. To delete the parameter, delete the corresponding line from the file.

Check services

Check the crond, sshd, and firewalld services. Specifically, ensure that crond and sshd are enabled, and firewalld is disabled.

Service	Description
crond	Recommended to be enabled. It is mainly used for setting up scheduled tasks and provides related O&M capabilities.
sshd	Recommended to be enabled. It is mainly used for encrypted login or data transmission and provides basic O&M capabilities.
firewalld	Recommended to be disabled, or add allow rules for each service in the OceanBase cluster to avoid firewall interception.

Manual modification method:

To enable a service: systemctl enable {serviceName} && systemctl start {serviceName}
To disable a service: systemctl stop {serviceName} && systemctl disable {serviceName}

Check Transparent Huge Pages (THP)

Check whether THP is disabled to ensure system performance stability.

Manual modification method: Set THP to never by running the command echo never > /sys/kernel/mm/transparent_hugepage/enabled. Additionally, to ensure the configuration persists after a reboot, add this command to the rc.local or systemd configuration.

Check Python version

OceanBase-related scripts depend on Python 2.7.x. Ensure that the Python version is compatible to guarantee the normal operation of the scripts.

Manual modification method: Install Python 2.7 using the package manager and verify that the version returned by the python -V command is 2.7.

Check if Swap is enabled

Check whether the Swap partition is enabled on the deployment machine. It is recommended to disable the Swap partition to avoid performance fluctuations.

Manual modification method: Execute swapoff/swapon -a and remove any Swap-related entries from /etc/fstab for persistence.

Check if the required commands exist

Ensure that the necessary toolkits for the operation and maintenance of OceanBase and OCP are available.

RPM package	Corresponding command	Description
net-tools	`netstat`, `arp`, `ether-wake`, `ifconfig`, `ipmaddr`, `iptunnel`, `mii-diag`, `mii-tool`, `nameif`, `plipconfig`, `route`, `slattach`	Check if the network-related command package is installed. These commands are essential for O&M and are recommended for installation.
mtr	`mtr`	A network testing tool essential for O&M. Recommended for installation.
tar	`tar`	A Linux packaging tool essential for O&M. Recommended for installation.
binutils	`addr2line`, `ar`, `as`, `c++filt`, `dwp`, `elfedit`, `gprof`, `ld`, `ld.bfd`, `ld.gold`, `nm`, `objcopy`, `objdump`, `ranlib`, `readelf`, `size`, `strings`, `strip`	A Linux binary toolset essential for O&M. Recommended for installation.
bind-utils	`delv`, `dig`, `host`, `mdig`, `nslookup`, `nsupdate`	A Linux DNS toolset. Recommended for installation.
curl	`curl`	A command-line text transfer tool essential for O&M. Recommended for installation.
nc	`nc`	A netcat network tool essential for O&M. Recommended for installation.
iproute	`arpd`, `bridge`, `cbq`, `ctstat`, `devlink`, `genl`, `ifcfg`, `ifstat`, `ip`, `lnstat`, `nstat`, `rdma`, `routef`, `routel`, `rtacct`, `rtmon`, `rtpr`, `rtstat`, `ss`, `tc`	A network management toolkit essential for O&M. Recommended for installation.

Manual modification method: Ensure that the network is functioning properly and the relevant software repositories are configured. Then, install the required packages using commands like yum/apt-get/zypper install.

Check if the required shared object (so) files exist

Check whether the following so files exist:

RPM package	Description
libaio	A dependency package for program execution
libatomic	A dependency package for program execution

Check if the admin user exists

Check whether the admin user is used to run OCP or OceanBase services.

Manual repair method: Use the useradd and groupadd commands to create an admin user with uid=500 and gid=500.

Check if SELinux is enabled

Check whether SELinux is enabled. SELinux uses Mandatory Access Control (MAC) to manage permissions for specific programs and resources. It is recommended to disable SELinux.

Manual repair method: Use the /usr/sbin/setenforce command to disable SELinux (set to 0) and modify the configuration in /etc/selinux/config to disabled.

Check the operating system version

Check whether the operating system version meets the following requirements:

Linux Operating System	Version	Server Architecture
Alibaba Cloud Linux	7.2 and above	x86_64 (including Hygon), ARM_64 (Kunpeng, Phytium)
Anolis OS	8.6 and above	x86_64 (including Hygon), ARM_64 (Kunpeng, Phytium)
KylinOS	V10	x86_64 (including Hygon), ARM_64 (Kunpeng, Phytium)
UOS	V20	x86_64 (including Hygon), ARM_64 (Kunpeng, Phytium)
NFSChina	4.0 and above	x86_64 (including Hygon), ARM_64 (Kunpeng, Phytium)
Inspur KOS	5.8	x86_64 (including Hygon), ARM_64 (Kunpeng, Phytium)
CentOS/Red Hat Enterprise Linux	7.2 and above Notice CentOS 8.X is not supported.	x86_64 (including Hygon), ARM_64 (Kunpeng, Phytium)
SUSE Enterprise Linux	12SP3 and above	x86_64 (including Hygon), ARM_64 (only Kunpeng and Phytium)
Debian	8.3 and above	x86_64 (including Hygon), ARM_64 (only Kunpeng and Phytium)

Manual repair method: Update the system version to meet the requirements.

Check ulimit information

ulimit is used to limit the resources consumed by processes started in the shell. The OBServer process involves several limits, including the maximum thread stack size (stack), the maximum number of file handles (open files), and the core file size (core file size). To avoid system limitations affecting the service, it is recommended to set the ulimit parameters to the following required configurations:

root soft nofile 655350
root hard nofile 655350
admin soft nofile 655350
admin hard nofile 655350
admin soft stack 10240
admin hard stack 10240
admin soft nproc 655360
admin hard nproc 655360
admin soft core unlimited
admin hard core unlimited

Manual fix: Create the /etc/security/limits.d/99-oceanbase_limits.conf file and write the above configuration into it, ensuring that this file has the highest priority.

Check clocksource information

Check whether the system configuration item clocksource meets the requirements. The recommended value is tsc. Other values may trigger performance issues.

Manual fix: Execute the echo tsc > /sys/devices/system/clocksource/clocksource0/current_clocksource command to set the clocksource value to tsc.

Check the default MTU value of the network card

Check the default MTU value of the network card. The recommended value is 1500. While excessively large or small MTU values will not affect system operation, they can impact network transmission efficiency and performance.

Manual fix:

Temporary solution
- Using the IP command: sudo ip link set dev eth0 mtu 1500 .
- Using the ifconfig command: sudo ifconfig eth0 mtu 1500 .
Permanent solution
- For Debian/Ubuntu: Edit the /etc/network/interfaces file and add the following content.
```
auto eth0
iface eth0 inet dhcp
mtu 1500
```
- For CentOS/RHEL: Edit the /etc/sysconfig/network-scripts/ifcfg-eth0 file and add or modify the following content.
```
MTU=1500
```

Check the NTP/Chrony time synchronization service

Check whether the host is synchronized with the time source. It is recommended to enable the NTP or Chrony service.