The current code enters the operational state after a singleton ring is
formed. The singleton ring increases the ring sequence number. This
can result in a problem that if the node is stopped and started before
the token timeout period, the newly restarted processor could end up
with the same
This problem is known and a patch is being worked on for it.
Regards
-steve
On Wed, 2007-10-24 at 10:59 +0900, Sanghun Kim wrote:
Hi, all.
I found IPC disconnect problem (due to POLLHUP) when initialize two
AIS services.
I created two threads(AMF CKPT) to perform healthchecking and
!
saClmDispatch, Return value: 9, should be [EMAIL PROTECTED] test]#
Regards
-steve
On Thu, 2007-10-25 at 16:05 -0700, Drew Moseley wrote:
Steven Dake wrote:
Send me a patch for the saftest and where you put the sleep and I'll
look if this is a valid scenario.
Patch below. I was able
Drew,
Submit a bug in the saftest issue tracker along with the patch:
http://sourceforge.net/tracker/?func=addgroup_id=117632atid=678679
Regards
-steve
On Tue, 2007-10-30 at 10:48 -0700, Drew Moseley wrote:
Steven Dake wrote:
The reason BAD_HANDLE is returned is because the handle has been
Thank you so much for reporting how to fix this problem. I have had
several people send me messages on this problem but none with solutions.
A faq entry will surely follow with this information.
Regards
-steve
On Thu, 2007-11-01 at 09:33 -0400, Ward, Timothy - SSD wrote:
High end Cisco
Forgive me for not responding earlier - I have been ill.
If you mean that you unload the NIC driver I have not thoroughly tested
such operation with openais. Instead it is designed to check that an
interface has been ifconfig ethX down within the downcheck interval. if
that happens, openais
merged.
thanks
On Fri, 2007-09-28 at 22:44 +1200, Angus Salkeld wrote:
Hi
This should sort out the make install for:
1)
make
make DESTDIR=/usr install
2)
make O=/tmp/ais-objects
make O=/tmp/ais-objects DESTDIR=/usr install
The patch is inline and attached.
-Angus
Makefile
This backports patch 1379 from trunk for the upcoming new IPC changes to
both whitetank and trunk.
Regards
-steve
Index: exec/cfg.c
===
--- exec/cfg.c (revision 1489)
+++ exec/cfg.c (working copy)
@@ -298,7 +298,7 @@
Ramesh,
I will fix the issue in 0.80.X branch and make a new release shortly. I
am fixing up some IPC bugs as the code goes through various QA
processes.
I would expect a release in a few weeks, but the problem should be
addressed in a day or two.
I'd recommend highly to stay on the 0.80.x
I'm not sure what you mean exactly. Do you have a test case?
Regards
-steve
On Thu, 2008-02-21 at 16:16 +0700, Trinh Van Thanh wrote:
Hi all,
In Checkpoint service have error, this occurs when execute creating/
writing a lot of checkpoints (Queue is full, do not writing to go down
Patch looks really good.
The two free operations are missing Puts on the msgHandle. Fix that up
and I'll go ahead and merge.
Regards
-steve
On Thu, 2008-02-21 at 11:40 -0600, Ryan O'Hara wrote:
Attached is an initial set of changes for the Message Service (B.03.01).
This patch includes
-Original Message-
From: Steven Dake [mailto:[EMAIL PROTECTED]
Sent: Tue 3/25/2008 5:27 PM
To: Sajeev Vallat
Cc: [EMAIL PROTECTED]
Subject: Re: [Openais] openais multicast address (node discovery issue)
Sajeev,
My guess is your switch is not configured for multicasting
Remove a double pthread_mutex_destroy. This is for the evs service.
Index: evs.c
===
--- evs.c (revision 1506)
+++ evs.c (working copy)
@@ -166,10 +166,7 @@
evs_inst-finalize = 1;
pthread_mutex_unlock (evs_inst-response_mutex);
Hat, Inc.
*
* All rights reserved.
*
- * Author: Steven Dake ([EMAIL PROTECTED])
+ * Author: Steven Dake ([EMAIL PROTECTED])
*
* This software licensed under BSD license, the text of which follows:
*
@@ -1029,7 +1029,7 @@
SaCkptCheckpointHandleT checkpointHandle
On Wed, 2008-03-26 at 12:36 -0500, David Teigland wrote:
On Wed, Mar 26, 2008 at 10:32:54AM -0500, David Teigland wrote:
A while back I drew this diagram to show what we were aiming to design, in
broad terms, for the next generation aisexec/cman config system:
In the condition that there are 3 nodes, and 2 nodes fail, and then 1
node is started, while that 1 node is executing the synchronization
recovery algorithm (for ckpt in this case), and then another node
starts, the checkpoint database can become corrupted because of an early
commit in the abort
On Fri, 2008-03-28 at 14:10 +1300, angus salkeld wrote:
Hi
Does anyone have a git mirror of openais?
If not:
Steve have you thought of migrating to git (or at least providing a git
mirror)?
1] It's nice getting credit for patches.
2] It's good to have offline history.
Thanks
Hardware requirements are a switch that is capable of multicasting and
atleast 1 node. More can be used if you desire. The kernel must be
compiled to support IGMP multicast (which the F8 kernel is). The
software in Fedora 8 is a bit dated. If you want to play with failover,
you should consider
Specify the full IP address openais should use in bindnetaddr. There is
a known issue with how netmasks are used to determine the ip addresses
to bind to which result in the inability to find the binding network
interface.
Regards
-steve
On Tue, 2008-04-01 at 13:44 +0530, Shridhar Sahukar
On Mon, 2008-04-21 at 09:20 +0530, pradheepa pradhi wrote:
Hi,
I have downloaded openais0.82.
I don't get the IP address of the node which have got joined.
For eg: My system IP address is 13.147.18.123. In
openais.conf file i have given
Lon
Good work! Thanks for the original patch.
I merged Fabio's rev of the patch.
I had originally thought forked processes inherited a copy of the forked
threads, but this is not the case.
Regards
-steve
On Thu, 2008-04-17 at 08:34 +0200, Fabio M. Di Nitto wrote:
Hi Lon
On Wed, 16 Apr
The patch looks good except for the extraneous changes to logsys_t2.c
which are unneeded.
Please remove that part of the patch and apply.
Regards
-steve
On Tue, 2008-04-22 at 10:50 +0100, Christine Caulfield wrote:
The logsys_config_priority_set() API call only changes the priority of
the
Patch merged.
Thanks Andrew
-steve
On Tue, 2008-04-22 at 13:05 +0200, Andrew Beekhof wrote:
___
Openais mailing list
Openais@lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/openais
The patch looks pretty good except the use of logsys in the objdb.
Any errors should be returned via the write interface function call, and
not logged via the logging system. If the user wants to log the errors,
they may do so.
The reason for this constraint is so that the object database may
On Mon, 2008-04-28 at 11:05 +0100, Christine Caulfield wrote:
This is a patch that adds some standalone code to libconfdb, so it can
examine and ultimately change, the configuration without aisexec running.
A small number of things I'm unsure about..
- it automatically falls back to
:
- *
* Copyright (c) 2006 MontaVista Software, Inc.
+ * Copyright (c) 2006-2007 Red Hat, Inc.
*
* All rights reserved.
*
- * Author: Steven Dake ([EMAIL PROTECTED])
+ * Author: Steven Dake ([EMAIL PROTECTED])
*
* This software licensed under BSD license, the text of which follows:
*
@@ -99,54
On Wed, 2008-04-30 at 14:14 +0100, Christine Caulfield wrote:
Steven Dake wrote:
On Mon, 2008-04-28 at 11:05 +0100, Christine Caulfield wrote:
This is a patch that adds some standalone code to libconfdb, so it can
examine and ultimately change, the configuration without aisexec running
.
*
- * Author: Steven Dake ([EMAIL PROTECTED])
+ * Author: Steven Dake ([EMAIL PROTECTED])
*
* This software licensed under BSD license, the text of which follows:
*
diff -uNr openais-trunk.orig/exec/aispoll.h openais-trunk.new/exec/aispoll.h
--- openais-trunk.orig/exec/aispoll.h 2007-06-04
See patch.
Not sure how this effects the srcdir functionality. If you have a
better patch which allows make from the exec directory please send it.
Regards
-steve
Index: exec/Makefile
===
--- exec/Makefile (revision 1531)
+++
Your change works and was merged. Hope this doesn't break srcdir :)
Regards
-steve
On Sat, 2008-05-03 at 08:10 +0200, Andrew Beekhof wrote:
On Fri, May 2, 2008 at 4:22 PM, Steven Dake [EMAIL PROTECTED] wrote:
See patch.
Not sure how this effects the srcdir functionality.
I'm guessing
Type up
this should take you to stack frame #1.
then print npats.
then print i
then print patternArray-patterns[i[.pattern
print edi-edi_patterns.patterns[i].pattern
print *edi
print patternArray-patterns[i].allocatedSize
print edi-edi_patterns.patterns[i].patternSize
This should give me some
of
the pattern field above. The allocatedsize should be some reasonable
variable not 3 gigabytes of memory :)
(gdb) print edi-edi_patterns.patterns[i].patternSize
$7 = 26
(gdb)
-Original Message-
From: Steven Dake [mailto:[EMAIL PROTECTED]
Sent: Tuesday, May 06, 2008 11:51 PM
Xinwei,
I apologize for not having reviewed your lock patches earlier. We are
working very hard (top priority) to release 0.83 and there is currently
an issue with it that is blocking several users which is consuming all
of my attention.
I promise I will review the lock patches as soon as 0.83
and kindly
say me the i will try
from first..
On 4/21/08, Steven Dake [EMAIL PROTECTED] wrote:
On Mon, 2008-04-21 at 09:20 +0530, pradheepa pradhi wrote:
Hi,
I have downloaded openais0.82.
I don't get the IP address of the node which have got joined
This is not possible with the current code.
What parameters do you want to make reconfigurable?
Regards
-steve
On Fri, 2008-05-09 at 11:25 +, Hegde, Ramesh (OpenCall) wrote:
Hello,
Is there a way to inform aisexec to reload the configuration
in /etc/ais/openais.conf without
I am pleased to announce the release of openais 0.83 (trunk-unstable).
This release adds new functionality useful for third party plugin
systems and continues our project's progression toward a split of the
openais APIS and core code (corosync) code base.
Features added in this version:
A new
currently the objdb interface is in a bit of a messy state. The reason
is the need for iteration contexts for multiple iterators on the same
object or different objects at around the same time.
This patch removes objdb_find_reset and objdb_find and replaces them
with object_find_create,
On Thu, 2008-05-15 at 08:33 +0100, Christine Caulfield wrote:
Steven Dake wrote:
currently the objdb interface is in a bit of a messy state. The reason
is the need for iteration contexts for multiple iterators on the same
object or different objects at around the same time.
This patch
During some testing of new development work in a package that uses the
checkpoint service, it was found that sometimes the global id used to
keep track of unlinked checkpoints was not synchronized.
This scenario would occur when a checkpoint was created on a node, then
it was unlinked and closed.
Amar,
I will produce a patch for your problem below.
I _highly_ recommend using the stable version of openais which is known
as the whitetank branch. It has undergone many thousands of hours of
testing and bug fixing efforts by the community and various companies on
top of 0.70 and is really
removes references to totemip.h from sync system.
Index: exec/logsys.c
===
--- exec/logsys.c (revision 1551)
+++ exec/logsys.c (working copy)
@@ -53,11 +53,7 @@
#include stdlib.h
#include pthread.h
-#include swab.h
#include
)
+++ sync.c (working copy)
@@ -1,13 +1,10 @@
/*
* Copyright (c) 2005-2006 MontaVista Software, Inc.
- * Copyright (c) 2006 Ericsson AB.
* Copyright (c) 2006-2007 Red Hat, Inc.
*
- * Author: Steven Dake ([EMAIL PROTECTED])
- * Author: Hans Feldt
- *
* All rights reserved.
*
+ * Author
removes totemip.h references from cpg code.
Index: exec/logsys.c
===
--- exec/logsys.c (revision 1551)
+++ exec/logsys.c (working copy)
@@ -53,11 +53,7 @@
#include stdlib.h
#include pthread.h
-#include swab.h
#include logsys.h
, Inc.
*
* All rights reserved.
*
* Author: Steven Dake ([EMAIL PROTECTED])
- * Mark Haverkamp ([EMAIL PROTECTED])
+ * Author: Mark Haverkamp ([EMAIL PROTECTED])
*
* This software licensed under BSD license, the text of which follows:
*
@@ -83,9 +83,6 @@
*
*/
-#ifndef
Patch attached
Index: Makefile
===
--- Makefile (revision 1552)
+++ Makefile (working copy)
@@ -217,7 +217,7 @@
# - fPIC rules required for service handler shared objects
../lcr/lcr_ifact.o: ../lcr/lcr_ifact.c
- $(CC) $(CFLAGS)
merged
thanks
-steve
On Thu, 2008-06-19 at 10:33 +0200, Fabio M. Di Nitto wrote:
Hi Steven,
the patch removes a lot of duplicate work that is already available in
logsys via logsys_facility_id_get that does the same (and better)
conversion.
Please apply.
Fabio
--
I'm going to
merged
thanks for the work!
-steve
On Thu, 2008-06-19 at 12:15 +0200, Fabio M. Di Nitto wrote:
Hi Steven,
the following patch removes a very confusing setting for logsys and fixes
the debug=on behavior as described in openais.conf.5.
The first problem is that we are mixing settings for
merged
thanks!
-steve
On Thu, 2008-06-19 at 16:28 +0200, Fabio M. Di Nitto wrote:
Hi Steven,
this patch allows logsys users to retrive configure information.
This is very useful if we want to set some defaults by
LOGSYS_DECLARE_SYSTEM and then change those settings from another piece of
As we talked about on irc config_set is proper place to do this and
adding another function isn't.
Regards
-steve
On Thu, 2008-06-19 at 16:44 +0200, Fabio M. Di Nitto wrote:
Hi Steven,
this patch covers a small corner case where we want to stop logging to
file forever.
with the actual
Merged!
Thanks for the patch - sorry for the delay.
regards
-steve
On Tue, 2008-05-06 at 15:18 +0800, Xinwei Hu wrote:
Hi all:
The attached patch is a try to clean up the test/testlck.c, which
has some obvious mistakes.
Please kindly help to review it. Thanks.
merged
thanks for the patch
regards
-steve
On Tue, 2008-05-06 at 15:22 +0800, Xinwei Hu wrote:
Hi all,
The attached patch is a try to improve the async version of lck
library.
I've tested it with test/testlck so far.
Please help to review it. Thanks.
Patch merged
thanks again!
-steve
On Tue, 2008-05-06 at 17:48 +0800, Xinwei Hu wrote:
Hi all,
att.
Please help to review. Thanks.
___
Openais mailing list
Openais@lists.linux-foundation.org
Shridhar,
Yes there has been a forward port of the sync and totem code from whitetank
to address these issues mailed to this list already. This, along with other
improvements, is going to be released today as openais 0.84 (unstable).
Regards
-steve
On Mon, Jun 23, 2008 at 5:43 PM, Shridhar
remove some unused variables.
___
Openais mailing list
Openais@lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/openais
I am pleased to announce the release of openais 0.84 (unstable) for
development work.
This release includes several stability improvements to the core
protocol, the synchronization engine, the distributed locking service,
and logsys systems. Ryan O'Hara also contributed significant feature
merged
thanks for the work!
-steve
On Mon, 2008-06-23 at 15:26 +0200, Fabio M. Di Nitto wrote:
Hi Steven,
this patch basically allows each subsystem to set its own logging
level as
configure option to override built time values.
Extra bonus it fixes the default setting for the
Dave,
Your patch looks reasonable but has a few issues which need to be
addressed.
It doesn't address the setting of logsys_subsys_id but defines it. I
want to avoid the situation where logsys_subsys_id is defined, but then
not set. What I suggest here is to set logsys_subsys_id to some known
looks good.
go ahead and commit.
man page changes would be nice addition :)
regards
-steve
On Tue, 2008-07-01 at 13:09 +0100, Christine Caulfield wrote:
It can be quite a nuisance that you can't query the internal state of
many openais plugins. Many times I have wanted to look into CPG to
looks good
committed
thanks for the patch dave
regards
-steve
On Wed, 2008-07-02 at 16:00 -0500, David Teigland wrote:
Problem with assert.h in previous patch fixed here.
___
Openais mailing list
Openais@lists.linux-foundation.org
Angus
This was unintentional
I'll fix it after the objdb reworking is finished.
Thanks for pointing it out.
Regards
-steve
On Thu, 2008-07-03 at 12:40 +1200, angus salkeld wrote:
In the forward port of a sync fix the function sync_request() in sync.c
got deleted, this is still referenced in
Could you be more specific?
Do you have a simple test case?
Are you trying to run 0.70 with 0.80.3 at the same time? Compatability
between versions is not supported.
Regards
-steve
On Thu, 2008-07-03 at 17:27 +0530, V.N. Raju wrote:
Hi,
I am having problems with sending and receiving events
Dave,
Looks good - applied
Thanks again!
-steve
On Thu, 2008-07-10 at 15:16 -0500, David Teigland wrote:
On Wed, Jul 02, 2008 at 09:32:45AM -0500, David Teigland wrote:
It doesn't address the setting of logsys_subsys_id but defines it. I
want to avoid the situation where
On Fri, 2008-07-11 at 09:20 +0200, Fabio M. Di Nitto wrote:
Hi Steven,
as discussed in a previous thread logsys_config_subsys_set does not allow
to set a lower priority (for example switching from LOG_LEVEL_DEBUG to
LOG_LEVEL_INFO).
This behaviour can be problematic if we ask people to
Looks good as per our discussions via irc.
committed
thanks for the patch!
Regards,
-steve
On Fri, 2008-07-11 at 12:55 +0200, Fabio M. Di Nitto wrote:
On Fri, 11 Jul 2008, Fabio M. Di Nitto wrote:
Hi Steven,
syslog should generally get only relevant messages for sysadmin tasks. All
merged
thanks
-steve
On Fri, 2008-07-11 at 15:12 +0200, Fabio M. Di Nitto wrote:
On Fri, 11 Jul 2008, Christine Caulfield wrote:
Fabio M. Di Nitto wrote:
Hi Steven,
as discussed in a previous thread logsys_config_subsys_set does not allow
to set a lower priority (for example
I am excited and pleased to announce the release of the Corosync Cluster
Engine 0.90 as well as a version of openais which plugs in on top of the
Corosync Cluster Engine.
This effort will allow us to multiply our development efforts and focus
our collective development on features we deem most
Angus,
We have discussed this for many hours in the past and the current issue
is this:
a) I will not accept dependencies on opneais on third party projects
(libxml2)
and
b) Fabbione doesn't want to maintain a minimalistic XML parser.
So until a maintainer for a minimialistic XML parser steps
merged.
thanks!
-steve
On Tue, 2008-07-29 at 08:25 +1200, angus salkeld wrote:
---
corosync/lib/Makefile|1 +
corosync/lib/confdb.c| 58
++
corosync/lib/sa-confdb.c | 12 +
corosync/lib/sa-confdb.h |1 +
4 files
.
+ *
+ * All rights reserved.
+ *
+ * Author: Steven Dake [EMAIL PROTECTED]
+ *
+ * This software licensed under BSD license, the text of which follows:
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following
The protocol is designed to work with an ifconfig of primary interface
down. Never tried lo, although this doesn't seem like a reasonable use
case to me.
Regards
-steve
On Wed, 2008-08-06 at 22:35 +0800, peter wrote:
HI,
The openais used PF_UNIX socket to communication between the client
On Wed, 2008-07-30 at 11:21 +0100, Christine Caulfield wrote:
2. external config modules are rather strange. If I include
corosync/engine/config.h (as I'm maybe not supposed to) then then
.readconfig routine gets called with an objdb* rather than a
corosync_api_v1* which the rest of the
On Fri, 2008-08-08 at 01:08 -0700, Joel Becker wrote:
On Thu, Jul 24, 2008 at 07:24:47AM -0700, Steven Dake wrote:
I am excited and pleased to announce the release of the Corosync Cluster
Engine 0.90 as well as a version of openais which plugs in on top of the
Corosync Cluster Engine
On Fri, 2008-08-08 at 11:36 +0200, Fabien Thomas wrote:
Hi all,
I've one question regarding the new corosync / openais organisation:
is there a way to build openais without installing corosync ?
i've tried with some -I but the problem is that the file hierarchy in
repository is not the
On Sat, 2008-08-09 at 12:15 +0200, Jerome Martin wrote:
Hi Steven, Andrew,
I identified the issue described in [Openais] I think I have hit a
bug,
but need confirmation. It was my fault, however I am not sure what
happens should not be handled more gracefully by openais and/or
pacemaker
I don't have a problem with this patch but it breaks shutdown.
Example:
Before patch (pressing ctrl-c)
Aug 11 8:21:18.601215 [SERV ] Unloading openais component:
corosync_confdb v0
Aug 11 8:21:18.601259 [SERV ] Unloading openais component: corosync_cpg
v0
Aug 11 8:21:18.601287 [SERV ]
On Wed, 2008-08-20 at 14:41 -0700, Joel Becker wrote:
On Wed, Aug 20, 2008 at 11:17:18AM -0700, Steven Dake wrote:
On Wed, 2008-08-20 at 18:52 +0200, Andrew Beekhof wrote:
We talked on irc and came to the conclusion that apparently it is
impossible to fix the kernel issue
good for commit
regards
-steve
On Tue, 2008-09-02 at 08:19 +0100, Christine Caulfield wrote:
This patch adds key_increment and key_decrement calls to the objdb and
confdb subsystems.
This is useful to provide atomic counters (ag handle numbers) for
long-running (though not persistent)
Alan,
Personally I would ignore those errors relating to sendmsg having an
uninitialized variable in the iovector. I think it is just a bug with
valgrind or something innocuous. I am not sure what causes it, but I am
aware they are there. If you have a suppression file or have a patch to
Angus
any update on when this error occurs?
Regards
-steve
On Wed, 2008-08-20 at 06:54 +1200, angus salkeld wrote:
Signed-off-by: Angus Salkeld [EMAIL PROTECTED]
---
exec/totemsrp.c | 12 +++-
1 files changed, 11 insertions(+), 1 deletions(-)
diff --git a/exec/totemsrp.c
the crash, but thought it was a good idea to
put in some basic protective measures (short of putting magic numbers
in the message).
-Angus
On Tue, 2008-09-02 at 13:08 -0700, Steven Dake wrote:
Angus
any update on when this error occurs?
Regards
-steve
On Wed, 2008-08-20 at 06:54
During overload conditions of the IPC system and the usage of the timer
system by ipc threads and totem threads simultaneously, a lock ordering
condition was found resulting in complete deadlock.
Patch attached to fix.
Index: exec/timer.c
When lib_exit returns -1 (indicating the totem queue is full and unable
to send new messages), the ipc system should allow the transmission of
totem messages in this condition. However, with the current code, a
global lock is held during this period, locking out the ability to send
totem messages
In rare circumstances under overload conditions, disconnecting an IPC
connection from IPC using the checkpoint service and thousands of
checkpoints can result in an assertion in the executive.
The solution is that the close routine should return -1 to the
lib_exit_fn function, which should then
Under overload conditions with thousands of checkpoints, and an exiting
process, it is possible that many simulatenous checkpoint expirations
will result in an assertion.
This patch resolves that issue.
Regards
-steve
Index: exec/ckpt.c
looks good for merge
Regards
-steve
On Mon, 2008-09-22 at 13:05 -0500, Ryan O'Hara wrote:
Here is a patch that fixes a couple bugs when adding/removing a queue
to/from a queue group.
When inserting a queue into a queue group, we must check that the
queue is not already a member of the
The ipc system is designed such that there is one thread that uses the
poll system call in a loop. When the outbound IPC kernel buffers are
filled up, the poll thread is interrupted via the pthread_kill library
function (which also sets POLLOUT) so that poll will allow fallthrough
when outbound
This patch fixes the evs service to properly deliver messages.
Regards
-steve
Index: test/testevs.c
===
--- test/testevs.c (revision 1651)
+++ test/testevs.c (working copy)
@@ -132,13 +132,13 @@
* Demonstrate evs_mcast_joined
Fabien,
Please give me an hour to look at why/how this is different then
whitetank. I'd like to maintain compatibility.
Thanks
-steve
On Tue, 2008-09-23 at 18:24 +0200, Fabien Thomas wrote:
Is there a problem to commit this fix in trunk ?
Another bonus question is why the 2nd port is not
Fabien,
The root of the problem is two fold.
First I ported all of the whitetank totem code to corosync minus the
totemconfig code. The totemconfig code does the htons operation instead
of it being done in lower level code as is done in the patch.
I think the patch is the proper solution but
) 2006-2007 Red Hat, Inc.
+ *
+ * All rights reserved.
+ *
+ * Author: Steven Dake ([EMAIL PROTECTED])
+ *
+ * This software licensed under BSD license, the text of which follows:
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided
CPG clients that run as normal priority can be starved by aisexec in
certain environments from being scheduled. This results in overflow of
the IPC connection disconnecting the client.
This patch forces the outbound thread to schedule at the same priority
as the rest of aisexec removing the
Shridhar,
Yes it has been tested a long time ago and worked great. I don't recall
the details of the configuration options but there is some important
information about that the node ID is routed or totem won't pick it up.
I know some others at Red Hat have recently investigated where we are in
Patch looks great with a few minor architecture issues.
I suggest merging the patch now, and we can address the arch issues in a
further patch.
Essentially how I would like to see this change is that the service
engine uses a quorum layer which is responsible for loading the
configured
*flt_data;
+int flt_data_size;
+
+#define FDHEAD_INDEX (flt_data_size)
+#define FDTAIL_INDEX (flt_data_size + 1)
+
/*
* Copyright (c) 2002-2004 MontaVista Software, Inc.
- * Copyright (c) 2006-2007 Red Hat, Inc.
+ * Copyright (c) 2006-2008 Red Hat, Inc.
*
* Author: Steven Dake ([EMAIL PROTECTED
On Thu, 2008-10-09 at 11:33 -0500, David Teigland wrote:
Wow that is a complicated solution. I though that simple and blackbox
went well together.
Completely agree, too complex. The logging code I copy into all the
daemons I write is at the opposite end of the spectrum; I doubt it's
===
--- services/pload.c (revision 0)
+++ services/pload.c (revision 0)
@@ -0,0 +1,329 @@
+/*
+ * Copyright (c) 2008 Red Hat, Inc.
+ *
+ * All rights reserved.
+ *
+ * Author: Steven Dake ([EMAIL PROTECTED])
+ *
+ * This software licensed under BSD
Feel free to merge everything but the totemnet and totemsrp bits. Those
patches are already in the logsysv2 tree.
Thanks
-steve
On Wed, 2008-10-22 at 09:19 +1300, angus salkeld wrote:
This should remove all the warnings from corosync.
---
exec/apidef.c |7 ++-
Patch looks good except for one compatibility issue.
syslog_level is changed to priority (which seems reasonable). Wouldn't
it be possible to make the syslog_level still have the same effect as
priority in the config file? I would prefer this.
Please wait to rebase until I have the logsysv2
good for commit
On Fri, 2008-10-24 at 10:38 +0100, Christine Caulfield wrote:
Currently if a reload operation fails, any subsystems listening for
notifications will get a RELOAD_NOTIFY_START callback but not a
RELOAD_NOTIFY_END callback.
This patch adds a RELOAD_NOTIFY_FAILED callback so
In some circumstances the received_flg of the commit token is not
properly configured. This patch addresses that issue by scanning the
commit token on the first rotation for nodes in the previous ring id
that are improperly set and setting them appropriately.
Regards
-steve
Index: totemsrp.c
1 - 100 of 1690 matches
Mail list logo