HP-UX Platform Oracle Boot instance encounters: ora-27154,ora-27300,ora-27301,ora-27302

Environment: HP-UX 11.31 + Oracle 11.2.0.4
Symptom: After you install Oracle on HPUX, you cannot start the instance after you configure the parameters by business requirements.
The error is as follows:

ORA-27154：post/wait create failedORA-27300：OS system dependent operation:semget failed with status: 28ORA-27301：OS failure message: No space left on deviceORA-27302：failure occurred at: sskgpcreates

• 1. Initial positioning
• 2. Verifying the conjecture
• 3. In-depth analysis

1. Initial positioning

Quickly determine that this is an instance can not start, that is, nomount this stage will not succeed, the first thought is that the parameter configuration is unreasonable.
After an attempt, the final discovery is that the processes parameter is set too high, causing the instance to fail to start, the current desired setting of 8000, and a test adjustment of 7000 to start successfully.

Go to MOS search HPUX platform This error does not find results, but there are other platform matching results:

• Database Startup fails with Ora-27300:os system dependent Operation:semget failed with status:28 (document ID 949468.1)
  
• Instance Startup fails with Error ora-27154,ora-27300,ora-27301,ora-27302 (document ID 314179.1)

The reasons for these documents are basically located in the SEM-related kernel parameter adjustment.

2. Verifying the conjecture

Find Hpux about the current settings for the SEM kernel parameters:

kctune -h -B semmni=4096kctune -h -B semmns=8192kctune -h -B semmnu=4092kctune -h -B semvmx=32767

These are the recommended setting values for Oracle's official documentation, but now it seems that the settings for these kernel parameters do not meet the processes values required by the user's business at this time.

A description of these kernel parameter values is searched online:

All indications are related to the SEM parameters, so try to modify the Semmni and Semmns parameters to twice times the value, i.e. 8192 and 16384.

kctune -h -B semmni=8192kctune -h -B semmns=16384--重启操作系统生效:shutdown -ry 0

After setting the processes to 8000 again, the instance can be started normally and the problem is resolved.

3. In-depth analysis

When the Semmni and Semmns parameter values are official document defaults, the process is set to 8000 per business requirement and the instance cannot be started. After you set the Semmni and Semmns parameter values to a value of twice, and then test the process to 16000, you also cannot start the instance.
It can be seen that the SEM signal volume and processes have some kind of association, and at this time to boot to the Nomount state, the actual and no user connection, indicating that the semaphore is pre-allocated, the following to specifically verify.
All of the following tests are starting the database to Nomount:

3.1 Set the processes value to the default value of 150
The results observed at this time IPCS:

[Email protected]        [/ORADATA/ORCL]IPCS-ASIPC status From/dev/kmem as of Fri June 1 16:57:15 2018T ID KEY MODE       OWNER GROUP CREATOR CGROUP nsems otime CTIME semaphores:s 0 0x4f1c06da--ra-------      root root root root 1 11:44:05 11:44:05s 1 0x411c01b6--ra-ra-ra-root      root root root 1 11:44:07 11:44:05s 2 0x4e0c0002--ra-ra-ra-root root Root root 2 11:44:07 11:44:05s 3 0x41203bb5--ra-ra-ra-root root root roo T 2 no-entry 11:44:05s 4 0x01090522--ra-r--r--root root root root 1 no-en                  Try 11:44:11s 8197 0x00a5c581--ra-------sfmdb users sfmdb users 11:44:13 11:44:13s               8198 0x00a5c582--ra-------sfmdb users sfmdb users, 11:44:13 11:44:13s  8199 0x00a5c583--ra-------sfmdb users sfmdb users, 11:44:13 11:44:13s 8200 0x00a5 c584--ra-------Sfmdb users sfmdb users 11:44:13 11:44:13s 8201 0x00a5c585--ra----      ---sfmdb users sfmdb users 11:44:13 11:44:13s 8202 0x00a5c586--ra-------Sfmdb     Users Sfmdb users 11:44:13 11:44:13s 8203 0x00a5c587--ra-------Sfmdb users      Sfmdb users 16:32:32 11:44:13s 0x4914942f--ra-r--r--root root root Root 1 11:44:32 11:44:32s 0x410c030b--ra-ra-ra-root root root root 1 11 : 44:33 11:44:33s 196622 0X5C23A1BC--ra-r-----Oracle DBA Oracle DBA 154 no-entry 16:47: 46

You can see that the value of Nsems is 154, at which point 150 processes can be satisfied.

3.2 Set processes value to 8000
The results observed at this time IPCS:

[Email protected]        [/ORADATA/ORCL]IPCS-ASIPC status From/dev/kmem as of Fri June 1 17:00:50 2018T ID KEY MODE       OWNER GROUP CREATOR CGROUP nsems otime CTIME semaphores:s 0 0x4f1c06da--ra-------      root root root root 1 11:44:05 11:44:05s 1 0x411c01b6--ra-ra-ra-root      root root root 1 11:44:07 11:44:05s 2 0x4e0c0002--ra-ra-ra-root root Root root 2 11:44:07 11:44:05s 3 0x41203bb5--ra-ra-ra-root root root roo T 2 no-entry 11:44:05s 4 0x01090522--ra-r--r--root root root root 1 no-en                  Try 11:44:11s 8197 0x00a5c581--ra-------sfmdb users sfmdb users 11:44:13 11:44:13s               8198 0x00a5c582--ra-------sfmdb users sfmdb users, 11:44:13 11:44:13s  8199 0x00a5c583--ra-------sfmdb users sfmdb users, 11:44:13 11:44:13s 8200 0x00a5 c584--ra-------Sfmdb users sfmdb users 11:44:13 11:44:13s 8201 0x00a5c585--ra----      ---sfmdb users sfmdb users 11:44:13 11:44:13s 8202 0x00a5c586--ra-------Sfmdb     Users Sfmdb users 11:44:13 11:44:13s 8203 0x00a5c587--ra-------Sfmdb users      Sfmdb users 16:32:32 11:44:13s 0x4914942f--ra-r--r--root root root Root 1 11:44:32 11:44:32s 0x410c030b--ra-ra-ra-root root root root 1 11 : 44:33 11:44:33s 229390 0X5C23A1BC--ra-r-----Oracle DBA Oracle DBA 2001 No-entry 17:00:                44s 49167 0x5c23a1bd--ra-r-----Oracle DBA Oracle DBA 2001 No-entry 17:00:44s 49168 0x5c23a1be--RA-R-----Oracle DBA Oracle DBA 2001 No-entry 17:00:44s 49169 0X5C23A1BF--ra-r-----       Oracle DBA Oracle DBA 2001 No-entry 17:00:44s 8210 0x5c23a1c0--ra-r-----Oracle DBA Oracle DBA 2001 No-entry 17:00:44

You can see that the Nsems value is 2001, with a total of 5 groups, at which point 8000 processes can be satisfied.

3.3 Set processes value to 16000
The results observed at this time IPCS:

[Email protected]        [/ORADATA/ORCL]IPCS-ASIPC status From/dev/kmem as of Fri June 1 17:10:22 2018T ID KEY MODE       OWNER GROUP CREATOR CGROUP nsems otime CTIME semaphores:s 0 0x4f1c06da--ra-------      root root root root 1 11:44:05 11:44:05s 1 0x411c01b6--ra-ra-ra-root      root root root 1 11:44:07 11:44:05s 2 0x4e0c0002--ra-ra-ra-root root Root root 2 11:44:07 11:44:05s 3 0x41203bb5--ra-ra-ra-root root root roo T 2 no-entry 11:44:05s 4 0x01090522--ra-r--r--root root root root 1 no-en                  Try 11:44:11s 8197 0x00a5c581--ra-------sfmdb users sfmdb users 11:44:13 11:44:13s               8198 0x00a5c582--ra-------sfmdb users sfmdb users, 11:44:13 11:44:13s  8199 0x00a5c583--ra-------sfmdb users sfmdb users, 11:44:13 11:44:13s 8200 0x00a5 c584--ra-------Sfmdb users sfmdb users 11:44:13 11:44:13s 8201 0x00a5c585--ra----      ---sfmdb users sfmdb users 11:44:13 11:44:13s 8202 0x00a5c586--ra-------Sfmdb     Users Sfmdb users 11:44:13 11:44:13s 8203 0x00a5c587--ra-------Sfmdb users      Sfmdb users 16:32:32 11:44:13s 0x4914942f--ra-r--r--root root root Root 1 11:44:32 11:44:32s 0x410c030b--ra-ra-ra-root root root root 1 11 : 44:33-11:44:33

As you can see, because Nomount error: ora-27154,ora-27300,ora-27301,ora-27302, the instance startup is unsuccessful, so there is no assignment for the Oracle user.

3.4 Set processes value to 14000
The results observed at this time IPCS:

[Email protected]        [/ORADATA/ORCL]IPCS-ASIPC status From/dev/kmem as of Fri June 1 17:11:53 2018T ID KEY MODE       OWNER GROUP CREATOR CGROUP nsems otime CTIME semaphores:s 0 0x4f1c06da--ra-------      root root root root 1 11:44:05 11:44:05s 1 0x411c01b6--ra-ra-ra-root      root root root 1 11:44:07 11:44:05s 2 0x4e0c0002--ra-ra-ra-root root Root root 2 11:44:07 11:44:05s 3 0x41203bb5--ra-ra-ra-root root root roo T 2 no-entry 11:44:05s 4 0x01090522--ra-r--r--root root root root 1 no-en                  Try 11:44:11s 8197 0x00a5c581--ra-------sfmdb users sfmdb users 11:44:13 11:44:13s               8198 0x00a5c582--ra-------sfmdb users sfmdb users, 11:44:13 11:44:13s  8199 0x00a5c583--ra-------sfmdb users sfmdb users, 11:44:13 11:44:13s 8200 0x00a5 c584--ra-------Sfmdb users sfmdb users 11:44:13 11:44:13s 8201 0x00a5c585--ra----      ---sfmdb users sfmdb users 11:44:13 11:44:13s 8202 0x00a5c586--ra-------Sfmdb     Users Sfmdb users 11:44:13 11:44:13s 8203 0x00a5c587--ra-------Sfmdb users      Sfmdb users 16:32:32 11:44:13s 0x4914942f--ra-r--r--root root root Root 1 11:44:32 11:44:32s 0x410c030b--ra-ra-ra-root root root root 1 11 : 44:33 11:44:33s 294926 0X5C23A1BC--ra-r-----Oracle DBA Oracle DBA 1750 No-entry 17:10:                55s 65551 0x5c23a1bd--ra-r-----Oracle DBA Oracle DBA 1750 No-entry 17:10:55s 65552 0x5c23a1be--RA-R-----Oracle DBA Oracle DBA 1750 no-entry 17:10:55s 65553 0X5C23A1BF--ra-r-----       Oracle DBA Oracle DBA 1750 no-entry 17:10:55s 24594 0x5c23a1c0--ra-r-----Oracle DBA Oracle DBA 1750 no-entry 17:10:55s 8211 0x5c23a1c1--ra-r-----Oracle DBA ORACL  E dba 1750 no-entry 17:10:55s 8212 0x5c23a1c2--ra-r-----Oracle DBA  1750 No-entry 17:10:55s 8213 0x5c23a1c3--ra-r-----Oracle DBA Oracle DBA 1750 No-entry 17:10:55s 0x5c23a1c4--ra-r-----Oracle DBA Oracle DBA 1750 No-entry 17:10:55

As you can see, the Nsems value is 1750, altogether 9 groups, at this time can satisfy 14000 processes.

Summary: Through this case, you can know that the IPCS look at the signal volume and process has a direct correlation. Consulting Our experts, there is still some kind of algorithm, but from the HPUX experiment, not all conform to the proposed algorithm rules, no longer delve into.

HP-UX Platform Oracle Boot instance encounters: ora-27154,ora-27300,ora-27301,ora-27302