Send Linux-ha-cvs mailing list submissions to
[email protected]
To subscribe or unsubscribe via the World Wide Web, visit
http://lists.community.tummy.com/mailman/listinfo/linux-ha-cvs
or, via email, send a message with subject or body 'help' to
[EMAIL PROTECTED]
You can reach the person managing the list at
[EMAIL PROTECTED]
When replying, please edit your Subject line so it is more specific
than "Re: Contents of Linux-ha-cvs digest..."
Today's Topics:
1. Linux-HA CVS: linux-ha by andrew from
([email protected])
2. Linux-HA CVS: lib by alan from ([email protected])
3. Linux-HA CVS: lrm by alan from ([email protected])
----------------------------------------------------------------------
Message: 1
Date: Thu, 9 Feb 2006 14:36:15 -0700 (MST)
From: [email protected]
Subject: [Linux-ha-cvs] Linux-HA CVS: linux-ha by andrew from
To: [EMAIL PROTECTED]
Message-ID: <[EMAIL PROTECTED]>
linux-ha CVS committal
Author : andrew
Host :
Module : linux-ha
Dir : linux-ha
Modified Files:
heartbeat.spec.in
Log Message:
List relevant open bugzilla entries
===================================================================
RCS file: /home/cvs/linux-ha/linux-ha/heartbeat.spec.in,v
retrieving revision 1.158
retrieving revision 1.159
diff -u -3 -r1.158 -r1.159
--- heartbeat.spec.in 9 Feb 2006 20:12:04 -0000 1.158
+++ heartbeat.spec.in 9 Feb 2006 21:36:14 -0000 1.159
@@ -1,4 +1,4 @@
-# $Id: heartbeat.spec.in,v 1.158 2006/02/09 20:12:04 andrew Exp $
+# $Id: heartbeat.spec.in,v 1.159 2006/02/09 21:36:14 andrew Exp $
# Workaround for a change in RPM 4.1.x; needs proper fixing soon!
##%define _unpackaged_files_terminate_build 0
@@ -159,6 +159,8 @@
+ Support migration and un-migration of resources and resource groups
+ Create crm_verify for checking configuration validity
+ Simplify the passing of XML to cibadmin
+ + Open bugs
+ + 1075, 1080, 1081, 1084
* Thu Sep 22 2005 Alan Robertson <[EMAIL PROTECTED]> (see doc/AUTHORS file)
+ Version 2.0.2 - small bug fix only release
------------------------------
Message: 2
Date: Thu, 9 Feb 2006 16:21:46 -0700 (MST)
From: [email protected]
Subject: [Linux-ha-cvs] Linux-HA CVS: lib by alan from
To: [EMAIL PROTECTED]
Message-ID: <[EMAIL PROTECTED]>
linux-ha CVS committal
Author : alan
Host :
Project : linux-ha
Module : lib
Dir : linux-ha/lib/clplumbing
Modified Files:
GSource.c
Log Message:
I removed the file descriptor from the poll loop.
I suspect this might have been the cause of the lrmd loop...
===================================================================
RCS file: /home/cvs/linux-ha/linux-ha/lib/clplumbing/GSource.c,v
retrieving revision 1.75
retrieving revision 1.76
diff -u -3 -r1.75 -r1.76
--- GSource.c 9 Feb 2006 18:48:09 -0000 1.75
+++ GSource.c 9 Feb 2006 23:21:45 -0000 1.76
@@ -1,4 +1,4 @@
-/* $Id: GSource.c,v 1.75 2006/02/09 18:48:09 alan Exp $ */
+/* $Id: GSource.c,v 1.76 2006/02/09 23:21:45 alan Exp $ */
/*
* Copyright (c) 2002 Alan Robertson <[EMAIL PROTECTED]>
*
@@ -267,6 +267,7 @@
return FALSE;
}
+ g_source_remove_poll(source, &fdp->gpfd);
g_source_remove(fdp->gsourceid);
fdp->gsourceid = 0;
g_source_unref(source);
------------------------------
Message: 3
Date: Thu, 9 Feb 2006 16:24:14 -0700 (MST)
From: [email protected]
Subject: [Linux-ha-cvs] Linux-HA CVS: lrm by alan from
To: [EMAIL PROTECTED]
Message-ID: <[EMAIL PROTECTED]>
linux-ha CVS committal
Author : alan
Host :
Project : linux-ha
Module : lrm
Dir : linux-ha/lrm/lrmd
Modified Files:
lrmd.c
Log Message:
Put in some more error messages and more comments better explaining the bugs
in the code.
===================================================================
RCS file: /home/cvs/linux-ha/linux-ha/lrm/lrmd/lrmd.c,v
retrieving revision 1.209
retrieving revision 1.210
diff -u -3 -r1.209 -r1.210
--- lrmd.c 9 Feb 2006 08:46:57 -0000 1.209
+++ lrmd.c 9 Feb 2006 23:24:13 -0000 1.210
@@ -1,4 +1,4 @@
-/* $Id: lrmd.c,v 1.209 2006/02/09 08:46:57 sunjd Exp $ */
+/* $Id: lrmd.c,v 1.210 2006/02/09 23:24:13 alan Exp $ */
/*
* Local Resource Manager Daemon
*
@@ -242,6 +242,8 @@
static gboolean on_op_timeout_expired(gpointer data);
static gboolean on_repeat_op_readytorun(gpointer data);
static void on_remove_client(gpointer user_data);
+static void destroy_pipe_ra_stderr(gpointer user_data);
+static void destroy_pipe_ra_stdout(gpointer user_data);
/* message handlers */
static int on_msg_register(lrmd_client_t* client, struct ha_msg* msg);
@@ -408,25 +410,53 @@
op->rsc_id = NULL;
op->exec_pid = 0;
+ /*
+ * This action prohibits capturing output after the
+ * process has died. FIXME!
+ * Note that really fixing this would require that
+ * the GSource become a semi-independent object
+ * whose lifetime is potentially greater than that
+ * of the operation which spawned it as described
+ * by bug 756.
+ * Note that if you simply remove the _del_fd()
+ * call below that bad things will happen, because
+ * the Gsource code refers directly to this operation
+ * structure which will have been freed in
+ * just a few lines below...
+ */
if ( NULL != op->ra_stdout_gsource) {
G_main_del_fd(op->ra_stdout_gsource);
op->ra_stdout_gsource = NULL;
}
+ /*
+ * This action prohibits capturing output after the
+ * process has died. FIXME!
+ * Note that really fixing this would require that
+ * the GSource become a semi-independent object
+ * whose lifetime is potentially greater than that
+ * of the operation which spawned it as described
+ * by bug 756.
+ * Note that if you simply remove the _del_fd()
+ * call below that bad things will happen, because
+ * the Gsource code refers directly to this operation
+ * structure which will have been freed in just a
+ * few lines below...
+ */
if ( NULL != op->ra_stderr_gsource) {
G_main_del_fd(op->ra_stderr_gsource);
op->ra_stderr_gsource = NULL;
}
- if (op->ra_stdout_fd != -1) {
+ if (op->ra_stdout_fd >= 0) {
close(op->ra_stdout_fd);
op->ra_stdout_fd = -1;
}
- if (op->ra_stderr_fd != -1) {
+ if (op->ra_stderr_fd >= 0) {
close(op->ra_stderr_fd);
op->ra_stderr_fd = -1;
}
- memset(op->first_line_ra_stdout, 0, sizeof(op->first_line_ra_stdout));
+ op->first_line_ra_stdout[0] = EOS;
lrmd_debug2(LOG_DEBUG, "lrmd_op_destroy: free the op whose address is
%p"
, op);
@@ -3040,10 +3070,10 @@
op->ra_stderr_fd = stderr_fd[0];
op->ra_stdout_gsource = G_main_add_fd(G_PRIORITY_HIGH
, stdout_fd[0], FALSE, handle_pipe_ra_stdout
- , op, NULL);
+ , op, destroy_pipe_ra_stdout);
op->ra_stderr_gsource = G_main_add_fd(G_PRIORITY_HIGH
, stderr_fd[0], FALSE, handle_pipe_ra_stderr
- , op, NULL);
+ , op, destroy_pipe_ra_stderr);
op->exec_pid = pid;
@@ -3304,7 +3334,26 @@
return rsc;
}
+static void
+destroy_pipe_ra_stdout(gpointer user_data)
+{
+ lrmd_op_t* op = (lrmd_op_t *)user_data;
+
+ if (op->ra_stdout_fd >= 0) {
+ close(op->ra_stdout_fd);
+ op->ra_stdout_fd = -1;
+ }
+}
+static void
+destroy_pipe_ra_stderr(gpointer user_data)
+{
+ lrmd_op_t* op = (lrmd_op_t *)user_data;
+ if (op->ra_stderr_fd >= 0) {
+ close(op->ra_stderr_fd);
+ op->ra_stderr_fd = -1;
+ }
+}
static gboolean
handle_pipe_ra_stdout(int fd, gpointer user_data)
{
@@ -3322,6 +3371,13 @@
fd = op->ra_stdout_fd;
}
+ if (fd < 0) {
+ lrmd_log(LOG_CRIT
+ , "%s:%d: Attempt to read from closed file descriptor."
+ , __FUNCTION__, __LINE__);
+ return FALSE;
+ }
+
if (0 != read_pipe(fd, &data, op)) {
if ( NULL != op->ra_stdout_gsource) {
G_main_del_fd(op->ra_stdout_gsource);
@@ -3346,9 +3402,14 @@
, op->rsc_id, op_type, data);
}
/*
- * This isn't quite correct yet - there is no idea of a "line"
+ * This code isn't correct - there is no idea of a "line"
* in the code as it's presently written...
- * but it probably works OK for now...
+ * It produces erratic and hard-to read messages in the logs.
+ * And possibly causes errors in interpreting 'heartbeat' style
+ * resource agents. FIXME
+ * And, it's not obvious how the meta-data (which is _many_
+ * lines long) is handled by this...
+ * I suspect multi-line metadata hasn't been tested :-(.
*/
if (op->first_line_ra_stdout[0] == EOS) {
strncpy(op->first_line_ra_stdout, data
@@ -3373,7 +3434,7 @@
G_main_del_fd(op->ra_stderr_gsource);
op->ra_stderr_gsource = NULL;
}
- if (op->ra_stderr_fd != -1) {
+ if (op->ra_stderr_fd >= 0) {
close(op->ra_stderr_fd);
op->ra_stderr_fd = -1;
}
@@ -3427,12 +3488,21 @@
if ((readlen < 0) && (errno !=0)) {
rc = -1;
- if (errno != EAGAIN) {
- cl_perror("%s::%d errno=%d, read"
- , __FUNCTION__, __LINE__, errno);
- }
+ switch (errno) {
+ default:
+ cl_perror("%s::%d fd %d errno=%d, read"
+ , __FUNCTION__, __LINE__
+ , fd, errno);
+ break;
- if (errno == EAGAIN) {
+ case EBADFD:
+ lrmd_log(LOG_CRIT
+ , "%s:%d"
+ " Attempt to read from closed file descriptor %d."
+ , __FUNCTION__, __LINE__, fd);
+ break;
+
+ case EAGAIN:
rsc = lookup_rsc(op->rsc_id);
op_type = ha_msg_value(op->msg, F_LRM_OP);
lrmd_log(LOG_ERR, "RA %s:%s:%s (process %d) failed to "
@@ -3550,6 +3620,10 @@
}
/*
* $Log: lrmd.c,v $
+ * Revision 1.210 2006/02/09 23:24:13 alan
+ * Put in some more error messages and more comments better explaining the bugs
+ * in the code.
+ *
* Revision 1.209 2006/02/09 08:46:57 sunjd
* Fix bug 1008 ( LRMd consumes too much CPU )
* But, donnot know the root casue yet. :-(
------------------------------
_______________________________________________
Linux-ha-cvs mailing list
[email protected]
http://lists.community.tummy.com/mailman/listinfo/linux-ha-cvs
End of Linux-ha-cvs Digest, Vol 27, Issue 41
********************************************
