| Summary | Orbix service raises an error in the PSS layer "errno is 12 (not enough space)" |
|---|---|
| Article Number | 13412 |
| Environment |
Orbix 6.x All Supported Operating Systems |
| Question/Problem Description | An Orbix service raises error in PSS layer: "errno is 12 (not enough space)" At runtime, the Orbix service raises the following error in its log file: "errno is 12 (not enough space)" A possible full error message for the IFR could read: ERROR: DB del failed; env is ifr_store, db is IRObjectPSHomeImpl:1.0, errno is 12 (Not enough space)
This error can also occur on any of the Orbix services, e.g in the CFR: E - DB del failed; env is cfr_store, db is IT_CFRStoreImpl/BindingHomeImpl:1.0, errno is 12 (Not enough space)
|
| Clarifying Information | |
| Error Message | |
| Defect/Enhancement Number | |
| Cause | The underlying Berkeley DB is running out of locks and lock objects. The default value is 1000. By increasing the maximum number of locks available to the Berkeley database, the problem can be overcome. |
| Resolution |
If the error is happening for the interface repository (IFR), change this variable in the configuration scope of your Orbix service: plugins:pss_db:envs:ifr_store:lk_max_lockers = "10000"; plugins:pss_db:envs:ifr_store:lk_max_objects = "10000"; The same variable might be set for other services. For the naming service, it would be: plugins:pss_db:envs:it_naming_store:lk_max_locks = "10000"; plugins:pss_db:envs:it_naming_store:lk_max_lockers = "10000"; plugins:pss_db:envs:it_naming_store:lk_max_objects = "10000"; |
| Workaround | |
| Notes | The Berkeley DB documentation page can be found at http://www.oracle.com/technetwork/database/berkeleydb/documentation/index.html and the documentation page for the Berkeley DB command line utilities (including db_stat) can be found at http://download.oracle.com/docs/cd/E17076_02/html/api_reference/C/utilities.html. To monitor the locks db_stat must be used, this can easily be built using the Berkeley source, just ensure you build the same version of db_stat as is the version of the database. After that the configuration will need to be changed in order to use db_stat to monitor the locks. The reason for this is that by default DB_PRIVATE is set for all Orbix databases. This flag basically only allows one process to access the database at a time, which means if the database and application are up and running then it is not possible to simultaneously get the information using db_stat. This is easy to overcome by setting the following in the Orbix configuration file - the example here is for naming database ..... plugins:pss_db:envs:it_naming_store:private = "false"; Now after re-starting the services the lock usage can be determined by using the following command (it is a requirement that db_stat is run from where it was built rather than copied away from there): /users/dublin/rd 29> /vob/tools_src/db-4.5.20/build_unix/db_stat -c -h /local4/users/rd/6.3.4/product/var/simples/dbs/naming_pivo |
| Attachment |
#Orbix
#KnowledgeDocs
#Orbix6
#database
