Hi,
here are two patches for imjournal dealing with module's termination.
They should resolve problems with duplicate messages and delayed
shutdown. More info is in the patches.
Tomas
>From 837c30918e909759e0164a3dfbcbd62e447b32c1 Mon Sep 17 00:00:00 2001
From: Tomas Heinrich <[email protected]>
Date: Thu, 23 May 2013 20:22:47 +0200
Subject: [PATCH 1/2] bugfix: imjournal's thread shouldn't be canceled
Because of cancelation, the correct possition in the journal was
lost. This resulted in duplicate messages appearing in the logs.
---
plugins/imjournal/imjournal.c | 10 ++++++++--
1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/plugins/imjournal/imjournal.c b/plugins/imjournal/imjournal.c
index 07ab446..aad7194 100755
--- a/plugins/imjournal/imjournal.c
+++ b/plugins/imjournal/imjournal.c
@@ -546,16 +546,22 @@ finalize_it:
ENDsetModCnf
+BEGINisCompatibleWithFeature
+CODESTARTisCompatibleWithFeature
+ if(eFeat == sFEATURENonCancelInputTermination)
+ iRet = RS_RET_OK;
+ENDisCompatibleWithFeature
+
+
BEGINqueryEtryPt
CODESTARTqueryEtryPt
CODEqueryEtryPt_STD_IMOD_QUERIES
CODEqueryEtryPt_STD_CONF2_QUERIES
CODEqueryEtryPt_STD_CONF2_setModCnf_QUERIES
+CODEqueryEtryPt_IsCompatibleWithFeature_IF_OMOD_QUERIES
ENDqueryEtryPt
-
-
BEGINmodInit()
CODESTARTmodInit
*ipIFVersProvided = CURR_MOD_IF_VERSION; /* we only support the current interface specification */
--
1.7.10.4
>From 1fab15333900a12c29750211290286ba6a5f319d Mon Sep 17 00:00:00 2001
From: Tomas Heinrich <[email protected]>
Date: Fri, 24 May 2013 04:14:35 +0200
Subject: [PATCH 2/2] bugfix: imjournal should respect termination request
sd_journal_wait() ignores EINTR and this behavior interferes with
module's termination signaling. Therefore, poll() is used instead.
---
plugins/imjournal/imjournal.c | 78 +++++++++++++++++++++++++++++++++++------
1 file changed, 67 insertions(+), 11 deletions(-)
diff --git a/plugins/imjournal/imjournal.c b/plugins/imjournal/imjournal.c
index aad7194..ae29154 100755
--- a/plugins/imjournal/imjournal.c
+++ b/plugins/imjournal/imjournal.c
@@ -30,6 +30,7 @@
#include <ctype.h>
#include <stdlib.h>
#include <time.h>
+#include <sys/poll.h>
#include <sys/socket.h>
#include <errno.h>
@@ -164,14 +165,6 @@ readjournal() {
int priority = 0;
int facility = 0;
- /* Get next journal message, if there is none, wait for next */
- while (sd_journal_next(j) == 0) {
- if (sd_journal_wait(j, (uint64_t) -1) < 0) {
- iRet = RS_RET_ERR;
- goto ret;
- }
- }
-
/* Get message text */
if (sd_journal_get_data(j, "MESSAGE", &get, &length) < 0) {
logmsgInternal(NO_ERRCODE, LOG_SYSLOG|LOG_INFO, (uchar *)"log message from journal doesn't have MESSAGE", 0);
@@ -388,14 +381,58 @@ persistJournalState () {
}
+/* Polls the journal for new messages. Similar to sd_journal_wait()
+ * except for the special handling of EINTR.
+ */
+static rsRetVal
+pollJournal()
+{
+ DEFiRet;
+ struct pollfd pollfd;
+ int r;
+
+ pollfd.fd = sd_journal_get_fd(j);
+ pollfd.events = sd_journal_get_events(j);
+ r = poll(&pollfd, 1, -1);
+ if (r == -1) {
+ if (errno == EINTR) {
+ /* EINTR is also received during termination
+ * so return now to check the term state.
+ */
+ ABORT_FINALIZE(RS_RET_OK);
+ } else {
+ char errStr[256];
+
+ rs_strerror_r(errno, errStr, sizeof(errStr));
+ errmsg.LogError(0, RS_RET_ERR,
+ "poll() failed: '%s'", errStr);
+ ABORT_FINALIZE(RS_RET_ERR);
+ }
+ }
+
+ assert(r == 1);
+
+ r = sd_journal_process(j);
+ if (r < 0) {
+ char errStr[256];
+
+ rs_strerror_r(errno, errStr, sizeof(errStr));
+ errmsg.LogError(0, RS_RET_ERR,
+ "sd_journal_process() failed: '%s'", errStr);
+ ABORT_FINALIZE(RS_RET_ERR);
+ }
+
+finalize_it:
+ RETiRet;
+}
+
+
BEGINrunInput
CODESTARTrunInput
/* this is an endless loop - it is terminated when the thread is
* signalled to do so. This, however, is handled by the framework,
* right into the sleep below.
*/
- int count = 0;
-
if (cs.stateFile[0] != '/') {
char *new_stateFile;
@@ -436,14 +473,32 @@ CODESTARTrunInput
}
while (glbl.GetGlobalInputTermState() == 0) {
+ int count = 0, r;
+
+ r = sd_journal_next(j);
+ if (r < 0) {
+ char errStr[256];
+
+ rs_strerror_r(errno, errStr, sizeof(errStr));
+ errmsg.LogError(0, RS_RET_ERR,
+ "sd_journal_next() failed: '%s'", errStr);
+ ABORT_FINALIZE(RS_RET_ERR);
+ }
+
+ if (r == 0) {
+ /* No new messages, wait for activity. */
+ CHKiRet(pollJournal());
+ continue;
+ }
+
CHKiRet(readjournal());
+ /* TODO: This could use some finer metric. */
count++;
if (count == cs.iPersistStateInterval) {
count = 0;
persistJournalState();
}
}
- persistJournalState();
finalize_it:
ENDrunInput
@@ -490,6 +545,7 @@ ENDwillRun
/* close journal */
BEGINafterRun
CODESTARTafterRun
+ persistJournalState();
sd_journal_close(j);
ENDafterRun
--
1.7.10.4
_______________________________________________
rsyslog mailing list
http://lists.adiscon.net/mailman/listinfo/rsyslog
http://www.rsyslog.com/professional-services/
What's up with rsyslog? Follow https://twitter.com/rgerhards
NOTE WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad of
sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you DON'T LIKE
THAT.
