Troubleshooting ORA-27300 ‘fork failed with status: 11’ on SLES12 (SUSE /Linux 7)

近日新装的一套ORACLE 12.2  RAC on SLES 12在使用srvctl start database 有时失败， alert log 中出现ORA-27300、ORA-27301、ORA-27302错误， 从错误不难看出是OS资源资源限制， 发现这可能以后会是个常见问题， 因为这是SLES 12的默认参数限制，而且ORACLE的安装文档和最佳实践中也未提到该参数（至少当前没有）。

alert log file

2018-11-19 09:29:03.940000 +08:00
Process P01E died, see its trace file
Process startup failed, error stack:
Errors in file /oracle/app/oracle/diag/rdbms/anbob/anbob1/trace/anbob1_psp0_32198.trc:
ORA-27300: OS system dependent operation:fork failed with status: 11
ORA-27301: OS failure message: Resource temporarily unavailable
ORA-27302: failure occurred at: skgpspawn3
2018-11-19 09:29:04.940000 +08:00
Process P01F died, see its trace file
Process startup failed, error stack:
Errors in file /oracle/app/oracle/diag/rdbms/anbob/anbob1/trace/anbob1_psp0_32198.trc:
ORA-27300: OS system dependent operation:fork failed with status: 11
ORA-27301: OS failure message: Resource temporarily unavailable
ORA-27302: failure occurred at: skgpspawn3

Note:

从当前的错误fork failed with status: 11可以大概猜测是最大进程数限制导致fork()进程时失败， 另外一处是psp0进程fork() Pnnn的进程。

PSP trace file

oracle@kdanbob01:/home/oracle> vi /oracle/app/oracle/diag/rdbms/anbob/anbob1/trace/anbob1_psp0_32198.trc
Trace file /oracle/app/oracle/diag/rdbms/anbob/anbob1/trace/anbob1_psp0_32198.trc
Oracle Database 12c Enterprise Edition Release 12.2.0.1.0 - 64bit Production
Build label:    RDBMS_12.2.0.1.0_LINUX.X64_170125
ORACLE_HOME:    /oracle/app/oracle/product/12.2.0/db_1
System name:    Linux
Node name:      kdanbob01
Release:        4.4.21-69-default
Version:        #1 SMP Tue Oct 25 10:58:20 UTC 2016 (9464f67)
Machine:        x86_64
Instance name: anbob1
Redo thread mounted by this instance: 0 
Oracle process number: 4
Unix process pid: 32198, image: oracle@kdanbob01 (PSP0)


*** 2018-11-19T09:15:26.133868+08:00
Process startup failed, error stack:
ORA-27300: OS system dependent operation:fork failed with status: 11
ORA-27301: OS failure message: Resource temporarily unavailable
ORA-27302: failure occurred at: skgpspawn3
OS - DIAGNOSTICS
----------------
loadavg : 42.40 27.56 11.09
Memory (Avail / Total) = 176425.47M / 515676.77M
Swap (Avail / Total) = 32768.00M /  32768.00M
Max user processes limits(s / h) =  65536 / 65536

oracle@kdanbob01:/home/oracle> ulimit -a

core file size (blocks, -c) 0

data seg size (kbytes, -d) unlimited

scheduling priority (-e) 0

file size (blocks, -f) unlimited

pending signals (-i) 2062629

max locked memory (kbytes, -l) unlimited

max memory size (kbytes, -m) unlimited

open files (-n) 65536

pipe size (512 bytes, -p) 8

POSIX message queues (bytes, -q) 819200

real-time priority (-r) 0

stack size (kbytes, -s) 10240

cpu time (seconds, -t) unlimited

max user processes (-u) 16384

virtual memory (kbytes, -v) unlimited

file locks (-x) unlimited

what’s psp Process？

The first process that will be started when we start instance is PSP process. This is called PROCESS SPAWNER. This process is introduced in 10g and is responsible for creating and managing other oracle backgroung processes.

Pnnn进程都知道是并行进程，查看并行相关的参数

SQL> show parameter parallel

PARAMETER_NAME                                               TYPE        VALUE
------------------------------------------------------------ ----------- ---------------------------------
containers_parallel_degree                                   integer     65535
fast_start_parallel_rollback                                 string      LOW
parallel_adaptive_multi_user                                 boolean     FALSE
parallel_degree_limit                                        string      CPU
parallel_degree_policy                                       string      MANUAL
parallel_execution_message_size                              integer     16384
parallel_force_local                                         boolean     TRUE
parallel_instance_group                                      string
parallel_max_servers                                         integer     60
parallel_min_percent                                         integer     0
parallel_min_servers                                         integer     60
parallel_min_time_threshold                                  string      AUTO
parallel_servers_target                                      integer     60
parallel_threads_per_cpu                                     integer     2
recovery_parallelism                                         integer     8
SQL> @pd containers_parallel_degree
Show all parameters and session values from x$ksppi/x$ksppcv...

      INDX I_HEX NAME                            VALUE                          DESCRIPTION
---------- ----- ------------------------------ ------------------------------ ---------------------------------------
      4749  128D containers_parallel_degree     65535                          Parallel degree for a CONTAINERS() query

Tip:

开始我以为是12cR2 新特性导致的该问题，containers_parallel_degree 当使用containers()查询或打开PDB时并行的限制值， 当前环境是个非CDB模式的数据库， 不过我个人还是建议在12.2 中使用多租户，哪怕是1PDBs, 在19c NO-CDB都是NO supperted。 如果只是因为是不想用CDB而多SCHEMA模式，我认为CDB优点更多:

言归正传回到刚才的问题，trace文件中提示系统max user process是65536， oracle当前用户环境限制是16384,  手动ps了一下当前用户进程根本没有那么多，只有一些后台进程不该存在这个问题， 在MOS中找到了一下找到找到的原因SLES 12: Database Startup Error with ORA-27300 ORA-27301 ORA-27303 While Starting using Srvctl (文档 ID 2340986.1)

From SLES12 onwards, systemd is used instead of initd and the OHASD server is only allowed to open a maximum of 512 tasks.

SYSTEMD从SLES12后替代了原来的OHASD服务， SYSTEMD是一个很受争议的东西，知乎上有篇说的不错，但是在 LINUX 上是主推。

在LINUX 6及以前的版本查’max user processes’ 使用 ‘ulimit -a’ ，但是以LINUX7(SLES12) 后可能需要查DefaultTasksMax （default value is 512）.

DefaultTasksMax   ==>systemd limited maximum number of tasks that may be created in the unit.This setting also effect maxpid value on OS.

$  >  systemctl status ohasd

● ohasd.service – LSB: Start and Stop Oracle High Availability Service

Loaded: loaded (/etc/init.d/ohasd; bad; vendor preset: disabled)

Active: active (exited) since Mon 2018-11-19 11:36:59 CST; 23h ago

Docs: man:systemd-sysv-generator(8)

Process: 12385 ExecStart=/etc/init.d/ohasd start (code=exited, status=0/SUCCESS)

Tasks: 726 (limit:

65535

) — 这个值在修改前应该是512

解决办法

Configure the value of DefaultTasksMax to 65535 in the file /etc/systemd/system.conf or or set the TasksMax value properly for the ohasd systemd service.

这里我们使用的解决方法是修改/etc/systemd/system.conf  把DefaultTasksMax改成 了65535，当然也可以直接改成’infinity’ 无限。

建议在SLSE 12或以后的版本，或LINUX 7等以后的版本时，先了解一下系统变化，至少在安装RAC时， 把DefaultTasksMax修改加入到安装方档中去， 可能Oracle 在以后的安装文档或最佳实践中会增加该内容。