Тонкая настройка Linux

Для тонкой настройки Linux:

1. Скопируйте содержимое конфигурационного файла в файл /etc/sysctl.d/sensors.conf.

   Примечание: содержимое конфигурационного файла для копирования

   #
       # Linux 3.10 tuning advisor for Sensors server.
       #
       # Note the guide is intended for large-scale deployments, therefore it is
       # assumed the servers are at least of 16 vCPU and 64GB RAM. The target number
       # of concurrent connections is 1M with request rate ~32K rps.
       #
       # The config should be saved to /etc/sysctl.d/sensors.conf and applied with the
       # following command:
       # > sysctl -p /etc/sysctl.d/sensors.conf
       #
       # Refer to http://man7.org/linux/man-pages/man5/sysctl.d.5.html for details.
       #
       #---------------------------------------------------------------------------
       # File Descriptors
       #
       # Ref:
       # - https://www.kernel.org/doc/Documentation/sysctl/fs.txt
       #
       # Set the maximum number of open file descriptors enforced at the kernel level.
       fs.file-max=10000000
       #
       # Max number of of file-handles a process can allocate.
       fs.nr_open=2000500
       #
       # Additionally, the limit should be updated with ulimit - 2M per-process file
       # descriptors (or use systemd LimitNOFILE=2000500):
       # > ulimit -n 2000500
       #
       #---------------------------------------------------------------------------
       # Networking
       #
       # Maximum number of packets, queued on the INPUT side, when the interface
       # receives packets faster than kernel can process them.
       net.core.netdev_max_backlog=32768
       #
       # tcp_max_syn_backlog sets the maximum number of half-open connections for which
       # the client has not yet sent an ACK response can be kept in the queue.
       #
       # Ref:
       # - http://lxr.linux.no/linux+v3.10/Documentation/networking/ip-sysctl.txt#L281.
       #
       net.ipv4.tcp_syncookies=1
       net.ipv4.tcp_max_syn_backlog=32768
       net.core.somaxconn=32768
       #
       # tcp_tw_reuse allows reusing sockets in TIME_WAIT state for new connections
       # when it is safe from protocol viewpoint. Default value is 0 (disabled).
       # The tcp_tw_reuse setting is particularly useful in environments where numerous
       # short connections are open and left in TIME_WAIT state, such as web servers.
       #
       # CAUTION: Don't use tcp_tw_recycle if the service is behind an LB or any NAT -
       # the server can't distinguish new incoming connections from different clients
       # behind the same NAT device.
       #
       # Ref:
       # - http://lxr.linux.no/linux+v3.10/Documentation/networking/ip-sysctl.txt#L459.
       #
       net.ipv4.tcp_tw_reuse=1
       #
       # Kernel buffers.
       #
       # OPTIONAL: For extra-large deployments only. Kernel regulates these buffers on
       # its own optimally in the vast majority of cases; rarely, it could be useful to
       # increase net.ipv4.tcp_mem.
       #
       # Ref:
       # - http://lxr.linux.no/linux+v3.10/Documentation/networking/ip-sysctl.txt#L380.
       # - http://lxr.linux.no/linux+v3.10/Documentation/networking/ip-sysctl.txt#L502.
       # - http://lxr.linux.no/linux+v3.10/Documentation/networking/ip-sysctl.txt#L296.
       #
       #net.core.rmem_max=16384
       #net.core.wmem_max=16384
       #net.ipv4.tcp_rmem=1024 4096 16384
       #net.ipv4.tcp_wmem=1024 4096 16384
       #net.ipv4.tcp_mem=1000000 1000000 1000000
       #
       # Outbound connections.
       #
       net.ipv4.ip_local_port_range=1024 65535
       net.ipv4.tcp_fin_timeout=30
       #
       #---------------------------------------------------------------------------
       # Miscellaneous
       #
       # Pre-allocate huge pages (16000 pages for 32GB RAM).
       #
       # OPTIONAL: For extra-large deployments only. The value is deployment specific
       # and should be adjusted in accordance to the actual server resources.
       # 
       # Better use of Translation-Lookaside Buffer (TLB) caches by the processor.
       # These caches contain virtual-to-physical address translations and have a small
       # number of entries with the most-recently used pages. Using huge pages of 2 MB
       # instead of the normal 4 KB pages, a TLB entry can handle much more memory,
       # thus making the CPU caching more efficient.
       #
       #vm.nr_hugepages=16000
       #

2. Выполните команду для применения конфигурации:

   sysctl -p /etc/sysctl.d/sensors.conf

3. Переопределите значение LimitNOFILE для systemd-юнита службы bzserver:

   systemctl edit bzserver

4. В открывшемся редакторе введите:

   [Service]
   LimitNOFILE=2000500

5. Примените изменения и перезапустите Сервер управления:

   systemctl daemon-reload
   systemctl restart bzserver

6. Убедитесь, что изменения вступили в силу:

   systemctl show -p LimitNOFILE bzserver
   cat /proc/$(pidof server)/limits | grep "Max open files"