mturk 2005/03/26 00:44:11
Modified: jk/native/common jk_lb_worker.c
Log:
In case ajp13 worker returns client error do not mark the worker
in error state. This operation is not recoverable, so return BAD_REQUEST
to web server.
Revision Changes Path
1.73 +50 -24 jakarta-tomcat-connectors/jk/native/common/jk_lb_worker.c
Index: jk_lb_worker.c
===================================================================
RCS file:
/home/cvs/jakarta-tomcat-connectors/jk/native/common/jk_lb_worker.c,v
retrieving revision 1.72
retrieving revision 1.73
diff -u -r1.72 -r1.73
--- jk_lb_worker.c 26 Feb 2005 07:16:54 -0000 1.72
+++ jk_lb_worker.c 26 Mar 2005 08:44:11 -0000 1.73
@@ -28,15 +28,16 @@
#include "jk_util.h"
#include "jk_worker.h"
#include "jk_lb_worker.h"
+#include "jk_ajp13.h"
#include "jk_mt.h"
#include "jk_shm.h"
/*
- * The load balancing code in this
+ * The load balancing code in this
*/
-/*
+/*
* Time to wait before retry...
*/
#define JK_WORKER_IN_ERROR(w) ((w)->in_error_state && !(w)->is_disabled &&
!(w)->is_busy)
@@ -66,9 +67,9 @@
if (strlen(id_start)) {
char *id_end;
id_start = jk_pool_strdup(s->pool, id_start);
- /*
+ /*
* The query string is not part of req_uri, however
- * to be on the safe side lets remove the trailing query
+ * to be on the safe side lets remove the trailing query
* string if appended...
*/
if ((id_end = strchr(id_start, '?')) != NULL) {
@@ -186,14 +187,14 @@
JK_TRACE_EXIT(l);
}
-static worker_record_t *find_by_session(lb_worker_t *p,
+static worker_record_t *find_by_session(lb_worker_t *p,
const char *name,
jk_logger_t *l)
{
worker_record_t *rc = NULL;
unsigned int i;
-
+
for (i = 0; i < p->num_of_workers; i++) {
if (strcmp(p->lb_workers[i].s->name, name) == 0) {
rc = &p->lb_workers[i];
@@ -214,7 +215,7 @@
size_t curmin = 0;
worker_record_t *candidate = NULL;
-
+
/* First try to see if we have available candidate */
for (i = 0; i < p->num_of_workers; i++) {
/* Skip all workers that are not member of domain */
@@ -258,7 +259,7 @@
unsigned int i;
int total_factor = 0;
worker_record_t *candidate = NULL;
-
+
/* First try to see if we have available candidate */
for (i = 0; i < p->num_of_workers; i++) {
/* If the worker is in error state run
@@ -286,14 +287,14 @@
return candidate;
}
-static worker_record_t *find_best_bytraffic(lb_worker_t *p,
+static worker_record_t *find_best_bytraffic(lb_worker_t *p,
jk_logger_t *l)
{
unsigned int i;
size_t mytraffic = 0;
size_t curmin = 0;
worker_record_t *candidate = NULL;
-
+
/* First try to see if we have available candidate */
for (i = 0; i < p->num_of_workers; i++) {
/* If the worker is in error state run
@@ -316,11 +317,11 @@
curmin = mytraffic;
}
}
- }
+ }
return candidate;
}
-static worker_record_t *find_bysession_route(lb_worker_t *p,
+static worker_record_t *find_bysession_route(lb_worker_t *p,
const char *name,
jk_logger_t *l)
{
@@ -379,7 +380,7 @@
{
worker_record_t *rc = NULL;
unsigned int i;
- const char *redirect = NULL;
+ const char *redirect = NULL;
for (i = 0; i < p->num_of_workers; i++) {
if (strlen(p->lb_workers[i].s->redirect)) {
@@ -522,6 +523,7 @@
lb_endpoint_t *p = e->endpoint_private;
int attempt = 0;
int num_of_workers = p->worker->num_of_workers;
+ worker_record_t *prec = NULL;
/* Set returned error to OK */
*is_error = JK_HTTP_OK;
@@ -539,12 +541,14 @@
worker_record_t *rec =
get_most_suitable_worker(p->worker, s, attempt++, l);
int rc;
-
- if (rec) {
+ /* Do not reuse previous worker, because
+ * that worker already failed.
+ */
+ if (rec && rec != prec) {
int is_service_error = JK_HTTP_OK;
- int service_ok = JK_FALSE;
+ int service_stat = JK_FALSE;
jk_endpoint_t *end = NULL;
-
+
s->jvm_route = rec->r;
rc = rec->w->get_endpoint(rec->w, &end, l);
@@ -561,7 +565,7 @@
/* Increment the number of workers serving request */
p->worker->s->busy++;
rec->s->busy++;
- service_ok = end->service(end, s, l, &is_service_error);
+ service_stat = end->service(end, s, l,
&is_service_error);
/* Update partial reads and writes if any */
rec->s->readed += end->rd;
rec->s->transferred += end->wr;
@@ -573,7 +577,7 @@
/* Decrement the busy worker count */
rec->s->busy--;
p->worker->s->busy--;
- if (service_ok) {
+ if (service_stat == JK_TRUE) {
rec->s->in_error_state = JK_FALSE;
rec->s->in_recovering = JK_FALSE;
rec->s->error_time = 0;
@@ -592,9 +596,10 @@
rec->s->name);
/* Decrement the worker count and try another worker */
--num_of_workers;
+ prec = rec;
continue;
}
- if (!service_ok) {
+ if (service_stat == JK_FALSE) {
/*
* Service failed !!!
*
@@ -623,6 +628,26 @@
"service failed, worker %s is in error state",
rec->s->name);
}
+ else if (service_stat == JK_CLIENT_ERROR) {
+ /*
+ * Clent error !!!
+ * Since this is bad request do not fail over.
+ */
+ is_service_error = JK_HTTP_BAD_REQUEST;
+ rec->s->errors++;
+ rec->s->in_error_state = JK_FALSE;
+ rec->s->in_recovering = JK_FALSE;
+ rec->s->error_time = 0;
+ *is_error = is_service_error;
+
+ jk_log(l, JK_LOG_INFO,
+ "unrecoverable error %d, request failed."
+ " Client failed in the middle of request,"
+ " we can't recover to another instance.",
+ is_service_error);
+ JK_TRACE_EXIT(l);
+ return JK_CLIENT_ERROR;
+ }
else {
/* If we can not get the endpoint from the worker
* that does not mean that the worker is in error
@@ -630,9 +655,8 @@
* We will try another worker.
* To prevent infinite loop decrement worker count;
*/
- --num_of_workers;
}
- /*
+ /*
* Error is recoverable by submitting the request to
* another worker... Lets try to do that.
*/
@@ -648,6 +672,8 @@
*is_error = JK_HTTP_SERVER_ERROR;
return JK_FALSE;
}
+ --num_of_workers;
+ prec = rec;
}
jk_log(l, JK_LOG_INFO,
"All tomcat instances are busy or in error state");
@@ -716,7 +742,7 @@
for (i = 0; i < num_of_workers; i++) {
p->lb_workers[i].s = jk_shm_alloc_worker(&p->p);
if (p->lb_workers[i].s == NULL) {
- jk_log(l, JK_LOG_ERROR,
+ jk_log(l, JK_LOG_ERROR,
"allocating worker record from shared memory");
JK_TRACE_EXIT(l);
return JK_FALSE;
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
