Bug #51610 [Wfx]: 1 second+ needed for OCI8 versus PDO/oci to open/close connection and exit

Edit report at http://bugs.php.net/bug.php?id=51610&edit=1 ID: 51610 Updated by: sixd@php.net Reported by: marioroy at verizon dot net Summary: 1 second+ needed for OCI8 versus PDO/oci to open/close connection and exit Status: Wont fix Type: Bug Package: OCI8 related Operating System: Linux & Snow Leopard PHP Version: 5.3SVN-2010-04-20 (snap) Assigned To: sixd New Comment: I wonder what issues the following patch would bring? Index: oci8.c =================================================================== --- oci8.c (revision 300858) +++ oci8.c (working copy) @@ -35,6 +35,7 @@ #include "php.h" #include "ext/standard/info.h" #include "php_ini.h" +#include "SAPI.h" #include "ext/standard/php_smart_str.h" #if HAVE_OCI8 @@ -3187,7 +3188,11 @@ PHP_OCI_CALL(OCIHandleFree, ((dvoid *) session_pool->err, OCI_HTYPE_ERROR)); } - if (session_pool->env) { + /* When PHP is invoked as a CLI then destruction of the + * Environment handle is left to the OS at PHP process shutdown. + */ + if (session_pool->env && !sapi_module.phpinfo_as_text) { PHP_OCI_CALL(OCIHandleFree, ((dvoid *) session_pool->env, OCI_HTYPE_ENV)); } Previous Comments: ------------------------------------------------------------------------ [2010-06-26 05:44:43] marioroy at verizon dot net I have applied this patch to oci8.c. It resolves the issue for me. I set use_spool to 0. Can we have use_spool enabled/disabled via an oci8.ini directive. Thanks. @@ -1711,7 +1693,7 @@ smart_str hashed_details = {0}; time_t timestamp; php_oci_spool *session_pool = NULL; - zend_bool use_spool = 1; /* Default is to use client-side session pool */ + zend_bool use_spool = 0; /* Default is to use client-side session pool */ zend_bool ping_done = 0; ------------------------------------------------------------------------ [2010-06-18 04:19:45] sixd@php.net The PHP OCI8 code changes in question have been in production for two years. They were introduced to support Database Resident Connection Pooling and FAN for scalability in web apps. The dominant use cases for SQL*Plus and PHP are very different. ------------------------------------------------------------------------ [2010-06-17 15:29:25] chris at leaflock dot net I filed a duplicate bug, but I noticed something unique; It doesn't happen with Oracle 10.2.0.3. I'm pretty unhappy about this, what is the benefit and/or requirement for this behavior, and why does it not happen in 10.2.0.3? Oracle's own SQLPlus doesn't display this behavior. I never thought I'd see the day when SQLPlus could do something faster for me than PHP. ------------------------------------------------------------------------ [2010-04-22 00:33:09] marioroy at verizon dot net Thank you for the feedback. ------------------------------------------------------------------------ [2010-04-21 18:31:34] sixd@php.net This is expected. There is a shutdown cost at process termination from closing a timer thread needed by the connection rearchitecture of OCI8 1.3. Most deployments either do not start and stop PHP processes frequently, or are not sensitive to the time between completing the work of the script and the termination of the process. It's possible that PDO_OCI may be changed in future and might get the same behaviour. ------------------------------------------------------------------------ The remainder of the comments for this report are too long. To view the rest of the comments, please view the bug report online at http://bugs.php.net/bug.php?id=51610 -- Edit this bug report at http://bugs.php.net/bug.php?id=51610&edit=1