Re: [Openais] Unable to start cluster (Pacemaker/Corosync)

Hi there, I am currently trying to configure Pacemaker/Corosync. I managed to install the required packages for the cluster configuration, however I could not start the cluster service. Based on the log file, there was an issue with the directory /var/lib/pacemaker/. I have tried some

[Openais] [PATCH] Ability to run as ais user

Attached patch solves problem with running corosync as ais user. Main problem was hidden in reading aisexec section. If this section exists in corosync.conf, everything works, but in other cases, main_config-uid/gid are initialized to 0 (so only root:root) can run corosync. Another problem, what

Re: [Openais] [PATCH] Ability to run as ais user

I hope last fix of that patch. Jim Meyering wrote: Jan Friesse wrote: Attached is patch with included suggested changes. Looks good. ... +The default is ais. one more: s/is/for each is/ Index: exec/mainconfig.c

[Openais] [PATCH] Replace sprintf with snprintf

Attached patch solves possible problems of overflow buffer, which can happened with sprintf. I tried to change only files where it makes sense (so basically, where in format is something like %s) and I hope I catch all cases. Index: test/sa_error.c

Re: [Openais] [PATCH] Ability to run as ais user

Steven, patch attached. Regards, Honza Steven Dake wrote: I already merged your previous patch. Can you rework the defines into a new patch? Thanks! Regards -steve On Wed, 2009-04-01 at 12:30 +0200, Jan Friesse wrote: Attached is patch with included suggested changes. Thanks Jim

Re: [Openais] [PATCH] Replace sprintf with snprintf

Jim, thanks for pointing bugs. Attached patch should be fine (I hope). Regards, Honza Jim Meyering wrote: Jan Friesse wrote: Attached patch solves possible problems of overflow buffer, which can happened with sprintf. I tried to change only files where it makes sense (so basically, where

[Openais] cpg - Corosync trunk - remove MESSAGE_REQ_CPG_GROUPS_GET ...

Patch to remove unused or by Steve unwanted functions message_handler_req_lib_cpg_groups_get and related things. commit ea17f0463b81378db123996ea8e33420c9c273ca Author: Jan Friesse jfrie...@redhat.com Date: Tue Apr 21 14:22:55 2009 +0200 CPG - Remove MESSAGE_REQ_CPG_GROUPS_GET. diff --git

Re: [Openais] [PATCH] logsys.c|h + main.c

a new internal but externally exported api that does the job for this special case without sprintfing in all cases or parsing the va args list in all cases (except when a va list is specified)? Regards -steve On Tue, 2009-03-31 at 17:04 +0200, Jan Friesse wrote: Attached patch solves

Re: [Openais] [PATCH] logsys.c|h + main.c

want to make it worse. Regards -steve On Thu, 2009-04-23 at 11:28 +0200, Jan Friesse wrote: Patch based on top of r2125. Performance impact should be one more call (shouldn't be so bad). Regards, Honza Steven Dake wrote: I can't tell, but it looks like this could have some

[Openais] [PATCH] openais/trunk: fixed evt service

Patch fixes evt service. Main problem is in hdb_handle_* functions, which expect 64bits handle number, but request has 32bits. So using hdb_nocheck_convert in service/evt.c makes test/testevt works again. Regards, Honza Index: services/evt.c

[Openais] [corosync trunk] CPG Fabbione problem

Fabbione, can you please try attached patch? I think (hope) it solves problem. Regards, Honza Index: lib/cpg.c === --- lib/cpg.c (revision 2158) +++ lib/cpg.c (working copy) @@ -355,7 +355,21 @@ goto error_put; break; }

Re: [Openais] [openais trunk] evt.c - Remove warnings

Sorry, bad patch included in last mail. Jan Friesse wrote: This patch fixes evt.c, so now no warnings are there. Regards, Honza Index: services/evt.c === --- services/evt.c (revision 1871) +++ services/evt.c (working copy

[Openais] [openais trunk] evt.c - Remove warnings

This patch fixes evt.c, so now no warnings are there. Regards, Honza Index: services/evt.c === --- services/evt.c (revision 1871) +++ services/evt.c (working copy) @@ -2583,12 +2583,6 @@ list_init(ecs-ecs_entry);

Re: [Openais] [openais trunk] evt.c - Remove warnings

to send it as an iovector to avoid the extra copy? If not, go ahead and merge, but the memcpy of event buffer data is somewhat detrimental to performance. Regards -steve On Thu, Apr 30, 2009 at 4:08 AM, Jan Friesse jfrie...@redhat.com wrote: Sorry, bad patch included in last mail. Jan

[Openais] [openais trunk] lib/msg.c handle NULL in queueName

Patch says it all. Regards, Honza diff --git a/trunk/lib/msg.c b/trunk/lib/msg.c index a631085..90473f2 100644 --- a/trunk/lib/msg.c +++ b/trunk/lib/msg.c @@ -380,6 +380,11 @@ saMsgQueueOpen ( /* DEBUG */ printf ([DEBUG]: saMsgQueueOpen\n); + if (queueName == NULL) { + error =

[Openais] [openais trunk] services/msg.c send MESSAGE_REQ_EXEC_MSG_QUEUEUNLINK instead of MESSAGE_REQ_EXEC_MSG_QUEUESTATUSGET

Patch says it all. Regards, Honza diff --git a/trunk/services/msg.c b/trunk/services/msg.c index 57a3cc5..13d6ae6 100644 --- a/trunk/services/msg.c +++ b/trunk/services/msg.c @@ -4857,7 +4857,7 @@ static void message_handler_req_lib_msg_queueunlink ( req_exec_msg_queueunlink.header.size =

[Openais] [openais trunk] lib/msg.c return SA_AIS_ERR_INVALID_PARAM instead of SA_AIS_ERR_INIT

Patch says it all. Regards, Honza diff --git a/trunk/lib/msg.c b/trunk/lib/msg.c index 90473f2..6d39452 100644 --- a/trunk/lib/msg.c +++ b/trunk/lib/msg.c @@ -501,7 +501,7 @@ saMsgQueueOpenAsync ( if ((openFlags SA_MSG_QUEUE_RECEIVE_CALLBACK)

[Openais] [openais trunk] lib/msg.c test openFlags SA_MSG_QUEUE_CREATE andcreationAttributes == NULL

Patch says it all. Regards, Honza diff --git a/trunk/lib/msg.c b/trunk/lib/msg.c index 571399d..6fed532 100644 --- a/trunk/lib/msg.c +++ b/trunk/lib/msg.c @@ -499,9 +499,14 @@ saMsgQueueOpenAsync ( goto error_exit; } + if ((openFlags SA_MSG_QUEUE_CREATE) + (creationAttributes

[Openais] Results of SAFTest for msg service after patches

Only for information. After apply all patches I send, you should get something like: SA Forum AIS Function Name Passing Tests Passing Percentage saMsgQueueUnlink19 of 19100.0% saMsgQueueOpenAsync 27 of 3675.0% saMsgQueueClose 10 of 1471.4%

[Openais] [PATCH 3/7] service/msg.c: message_handler_req_exec_msg_messageget test if msg is NULL

This solves another segfault in msg service (and corosync). --- trunk/services/msg.c | 10 +- 1 files changed, 5 insertions(+), 5 deletions(-) diff --git a/trunk/services/msg.c b/trunk/services/msg.c index 00f339d..4e91202 100644 --- a/trunk/services/msg.c +++ b/trunk/services/msg.c @@

[Openais] [PATCH 4/7] service/msg.c: Handle opening existing queue with different attributes

See P33-13:P33-14 --- trunk/services/msg.c | 23 +++ 1 files changed, 23 insertions(+), 0 deletions(-) diff --git a/trunk/services/msg.c b/trunk/services/msg.c index 4e91202..12ea394 100644 --- a/trunk/services/msg.c +++ b/trunk/services/msg.c @@ -2552,6 +2552,17 @@ static

[Openais] [PATCH 2/7] services/msg.c: message_handler_req_exec_msg_messagesendasync - test error and then queue

This solves one of segfaults of msg service (and whole corosync) --- trunk/services/msg.c |2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/trunk/services/msg.c b/trunk/services/msg.c index 17a50f9..00f339d 100644 --- a/trunk/services/msg.c +++ b/trunk/services/msg.c @@

[Openais] [PATCH 5/7] lib/msg.c: Handle creationFlags and coroipcc_msg_send_reply_receive errors

--- trunk/lib/msg.c | 49 ++--- 1 files changed, 42 insertions(+), 7 deletions(-) diff --git a/trunk/lib/msg.c b/trunk/lib/msg.c index b12eb2b..cc0a9b5 100644 --- a/trunk/lib/msg.c +++ b/trunk/lib/msg.c @@ -385,6 +385,24 @@ saMsgQueueOpen (

[Openais] [PATCH 7/7] lib/msg.c: Handle error codes from coroipc and similar functions

--- trunk/lib/msg.c | 75 -- 1 files changed, 61 insertions(+), 14 deletions(-) diff --git a/trunk/lib/msg.c b/trunk/lib/msg.c index cc0a9b5..2c688f9 100644 --- a/trunk/lib/msg.c +++ b/trunk/lib/msg.c @@ -665,6 +665,9 @@ saMsgQueueClose (

[Openais] [PATCH 6/7] service/msg.c: Return SA_AIS_ERR_NOT_EXIST as specification says

See P35 1-2 --- trunk/services/msg.c | 98 +++-- 1 files changed, 54 insertions(+), 44 deletions(-) diff --git a/trunk/services/msg.c b/trunk/services/msg.c index 12ea394..486fe9e 100644 --- a/trunk/services/msg.c +++ b/trunk/services/msg.c @@

[Openais] [PATCH] service/msg: Remove test of retentionTime

--- trunk/services/msg.c |2 -- 1 files changed, 0 insertions(+), 2 deletions(-) diff --git a/trunk/services/msg.c b/trunk/services/msg.c index 4d3615b..cfbed22 100644 --- a/trunk/services/msg.c +++ b/trunk/services/msg.c @@ -2495,7 +2495,6 @@ static void

[Openais] [PATCH] lib/msg.c: Better test of openFlags.

--- trunk/lib/msg.c | 10 ++ 1 files changed, 10 insertions(+), 0 deletions(-) diff --git a/trunk/lib/msg.c b/trunk/lib/msg.c index 2c688f9..56ac05a 100644 --- a/trunk/lib/msg.c +++ b/trunk/lib/msg.c @@ -385,6 +385,11 @@ saMsgQueueOpen ( goto error_exit; } +

[Openais] [PATCH] Test queueSize only if we wan't to create queue.

--- trunk/services/msg.c | 36 1 files changed, 20 insertions(+), 16 deletions(-) diff --git a/trunk/services/msg.c b/trunk/services/msg.c index dfcf4fa..ac04b00 100644 --- a/trunk/services/msg.c +++ b/trunk/services/msg.c @@ -2507,17 +2507,19 @@ static

[Openais] [PATCH] Check create_attrs only if it make sense (openFlags has SA_MSG_QUEUE_CREATE)

--- trunk/services/msg.c | 36 1 files changed, 20 insertions(+), 16 deletions(-) diff --git a/trunk/services/msg.c b/trunk/services/msg.c index cfbed22..dfcf4fa 100644 --- a/trunk/services/msg.c +++ b/trunk/services/msg.c @@ -2494,14 +2494,16 @@ static

[Openais] [PATCH] lib/msg.c: Make sure, that close really remove handle from db

--- trunk/lib/msg.c |5 + 1 files changed, 5 insertions(+), 0 deletions(-) diff --git a/trunk/lib/msg.c b/trunk/lib/msg.c index 33a81af..df2655d 100644 --- a/trunk/lib/msg.c +++ b/trunk/lib/msg.c @@ -684,6 +684,11 @@ saMsgQueueClose ( goto error_put; /* ! */ } +

[Openais] [corosync trunk] [patch] Handle EOL in coroparse.c

This is side effect of gid-uid patch I'm working on. Patch handles situations, where: - File is bad - no error message is written (because it's replaced AFTER parse_section) - No newline is in the end of file (yes I know, it's not proper text file, but ...) - Handle obscure OS, where line doesn't

Re: [Openais] [corosync trunk] [patch] Handle EOL in coroparse.c

Sorry, I forgot include patch. Jan Friesse wrote: This is side effect of gid-uid patch I'm working on. Patch handles situations, where: - File is bad - no error message is written (because it's replaced AFTER parse_section) - No newline is in the end of file (yes I know, it's not proper

[Openais] [corosync trunk] [patch] Support for reading configuration files from directory

it?) Regards, Honza commit 20090d706aa8e000d06449f0a65c85286207f601 Author: Jan Friesse jfrie...@redhat.com Date: Tue May 12 18:07:33 2009 +0200 coroparse: Support for reading configuration files from /etc/ais/security diff --git a/trunk/exec/coroparse.c b/trunk/exec/coroparse.c index

[Openais] [corosync trunk] [patch] Initial support for security (uid-gid)

Attached is first version of support for multiple security items (uid-gid). First question what I have. I'm currently testing uid and gid as a pair, so user process must have gid and uid (not only uid or gid). Is that correct, or you will rather see something, what will check uid OR gid? (From my

Re: [Openais] [corosync trunk] [patch] Initial support for security (uid-gid)

should be valid, not requiring an and operation. On Wed, 2009-05-13 at 18:21 +0200, Jan Friesse wrote: Attached is first version of support for multiple security items (uid-gid). First question what I have. I'm currently testing uid and gid as a pair, so user process must have gid and uid

Re: [Openais] [corosync trunk] [patch] Initial support for security (uid-gid)

Fabio, Fabio M. Di Nitto wrote: Hi Jan, I have few comments... On Thu, 2009-05-14 at 11:49 +0200, Jan Friesse wrote: differences between files attachment (corosync-support-for-uidgid-try2.patch) diff --git a/trunk/exec/main.c b/trunk/exec/main.c index db22e96..2b4 100644

Re: [Openais] [corosync trunk] [patch] Handle EOL in coroparse.c

Patched version of patch. Can you please ACK? Regards, Honza Jan Friesse wrote: Sorry, I forgot include patch. Jan Friesse wrote: This is side effect of gid-uid patch I'm working on. Patch handles situations, where: - File is bad - no error message is written (because it's replaced

[Openais] [corosync] [patch] Support for uidgid - another try

I hope I included all needed. Regards, Honza commit 2b29ee8c3fd049e56a72cac8f42182b450b3dc6f Author: Jan Friesse jfrie...@redhat.com Date: Mon May 18 14:56:18 2009 +0200 Support for uidgid.d feature. diff --git a/trunk/exec/coroparse.c b/trunk/exec/coroparse.c index 830b3ca..261c722

[Openais] [openais +corosync trunk] [patch] Support for corosync parser only and read configuration from /etc/corosync/....

With LCR patch everything seems working, so we can say goodbye openaisparser. Regards, Honza commit 2c03e74d87038cf2ebd0b73228f8153986131b6f Author: Jan Friesse jfrie...@redhat.com Date: Thu May 14 16:02:33 2009 +0200 Support for corosync parser only diff --git a/trunk/Makefile.am b

[Openais] [corosync + openais] [patch] Dispatch return bad handle - proposed solution

Hi, attached are proposed solution to *dispatch* functions, which returns CS_ERR_BAD_HANDLE (AIS_ERR_BAD_HANDLE (9)). David, can you please test them, and give results? Anyway, this patches helps to SAF test, to pass CLM, EVT 100% and helps checkpoint to get one more point, so from my point of

[Openais] [openais] [patch] msg: Handle multiple list del

Hi, included patch solves multiple list_del in saMsgQueueClose, which leads to segfault. Regards, Honza diff --git a/trunk/lib/msg.c b/trunk/lib/msg.c index 1c20119..689ffbc 100644 --- a/trunk/lib/msg.c +++ b/trunk/lib/msg.c @@ -78,6 +78,7 @@ struct queueInstance { SaMsgQueueHandleT

Re: [Openais] [openais] [patch] msg: Handle multiple list del

, 2009 at 03:43:09PM +0200, Jan Friesse wrote: Hi, included patch solves multiple list_del in saMsgQueueClose, which leads to segfault. Regards, Honza diff --git a/trunk/lib/msg.c b/trunk/lib/msg.c index 1c20119..689ffbc 100644 --- a/trunk/lib/msg.c +++ b/trunk/lib/msg.c @@ -78,6 +78,7

[Openais] [openais] [patch] Support for openaisserviceenable(stable|experimental). Stable doesn't include AMF.

SSIA https://bugzilla.redhat.com/show_bug.cgi?id=501695 Regards, Honza commit c9751a89328cfeb653a43387c70f6009184b3fd0 Author: Jan Friesse jfrie...@redhat.com Date: Wed May 20 17:57:38 2009 +0200 Support for openaisserviceenable(stable|experimental). Stable doesn't include AMF. diff

[Openais] [corosync] [patch] - ckpt solution - Change of Makefile.am

Hi, included is patch for Makefile.am of corosync, so coroipcc.o is no longer included in lib... directly, but rather *.so is a dependency, so ipc_hdb is no longer in multiple *.so and multiple times in binary what causes problem. Should solve https://bugzilla.redhat.com/show_bug.cgi?id=499918.

Re: [Openais] [corosync] [patch] - ckpt solution - Change of Makefile.am - try2

Attached is second version of patch. I tested this 300% more harder then previous patch, so I hope it will work. Regards, Honza David Teigland wrote: On Wed, May 27, 2009 at 04:15:52PM +0200, Jan Friesse wrote: Hi, included is patch for Makefile.am of corosync, so coroipcc.o is no longer

[Openais] [lck 0/7] LCK service patches

]: saLckResourceOpen [DEBUG]: saLckResourceLock [DEBUG]: saLckResourceLockAsync And lockAsync will never end. I didn't have time to figure our more deeply where is problem, so maybe you know, why it happend. Jan Friesse (7): Remove some warnings Sizeof should be structure and not pointer Test lockMode

[Openais] [lck 2/7] Sizeof should be structure and not pointer

--- trunk/services/lck.c |4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/trunk/services/lck.c b/trunk/services/lck.c index 099835c..383e862 100644 --- a/trunk/services/lck.c +++ b/trunk/services/lck.c @@ -2042,7 +2042,7 @@ error_exit: * Create

[Openais] [lck 4/7] Test resourceFlags in saResourceOpen(Async)

--- trunk/lib/lck.c | 10 ++ 1 files changed, 10 insertions(+), 0 deletions(-) diff --git a/trunk/lib/lck.c b/trunk/lib/lck.c index 1bb5420..cecf2a9 100644 --- a/trunk/lib/lck.c +++ b/trunk/lib/lck.c @@ -475,6 +475,11 @@ saLckResourceOpen ( goto error_exit; }

[Openais] [lck 5/7] Test validity of handle in *resourceLock(Async)

This patch also split lck_resourcelock_response_send to two functions, because it looks like easy way how to avoid code duplication. --- trunk/services/lck.c | 56 ++--- 1 files changed, 39 insertions(+), 17 deletions(-) diff --git

[Openais] [lck 6/7] Remove some compiler warnings.

It looks like really useful warning is only in header file, because there were one ; character more, then should be. --- trunk/include/ipc_lck.h |2 +- trunk/services/lck.c|9 + 2 files changed, 6 insertions(+), 5 deletions(-) diff --git a/trunk/include/ipc_lck.h

[Openais] [lck 7/7] In saLckOptionCheck, test lckOptions to NULL

--- trunk/lib/lck.c |4 1 files changed, 4 insertions(+), 0 deletions(-) diff --git a/trunk/lib/lck.c b/trunk/lib/lck.c index cecf2a9..b881651 100644 --- a/trunk/lib/lck.c +++ b/trunk/lib/lck.c @@ -263,6 +263,10 @@ saLckOptionCheck ( struct lckInstance *lckInstance;

Re: [Openais] [lck 2/7] Sizeof should be structure and not pointer

Of course we can. Ryan O'Hara wrote: This patch is correct, but what I intended was to use sizeof (struct resource_instance). Can we do that instead? On Tue, Jun 02, 2009 at 05:34:04PM +0200, Jan Friesse wrote: --- trunk/services/lck.c |4 ++-- 1 files changed, 2 insertions(+), 2

Re: [Openais] [lck 6/7] Remove some compiler warnings.

ok Ryan O'Hara wrote: Actually, can we just #ifdef out the print routines? They are useful for debugging while we work out the bugs. Just easier to remove ifdef here and there. On Tue, Jun 02, 2009 at 05:34:08PM +0200, Jan Friesse wrote: It looks like really useful warning is only

Re: [Openais] [lck 5/7] Test validity of handle in *resourceLock(Async)

it in old function) to send IPC reply. Ryan O'Hara wrote: NACK. What is the purpose of introducing a new function? On Tue, Jun 02, 2009 at 05:34:07PM +0200, Jan Friesse wrote: This patch also split lck_resourcelock_response_send to two functions, because it looks like easy way how to avoid

Re: [Openais] [lck 5/7] Test validity of handle in *resourceLock(Async)

Ryan O'Hara wrote: On Tue, Jun 02, 2009 at 05:58:46PM +0200, Jan Friesse wrote: Main problem is, that sometimes message_handler_req_lib_lck_resourcelock(async) ask to deleted handle. So this patch test result of hdb_get... and if it is not 0, it should return error to caller - ipc send. So

Re: [Openais] [PATCH] don't let a NUL input byte cause memory corruption (line[-1] = 0; )

Jim, ACK from me, but only amfutil does make sense. We will throw away openaisparser and replace it by coroparse, which doesn't have this problem, it's able to handle EOLs like \r, \n, \r\n, and handle lines without newline. At least I hope this is still plan, before 1.0. Regards, Honza Jim

[Openais] [lck 1/4] Return SA_AIS_ERR_NOT_EXIST in saLckResourceUnlock

This is very controversal, and I'm still not sure about correctnes of this. Becuase: - AIS B.03.01 don't have ERR_BAD_HANDLE as a posibility of returned code and can return only SA_AIS_ERR_NOT_EXIST - AIS B.01.01 can return ERR_BAD_HANDLE and ERR_NOT_EXISTS And because SAF Test is written for

[Openais] [lck 2/4] Test lockId to NULL

--- trunk/lib/lck.c | 10 ++ 1 files changed, 10 insertions(+), 0 deletions(-) diff --git a/trunk/lib/lck.c b/trunk/lib/lck.c index 5d2dbce..6a43447 100644 --- a/trunk/lib/lck.c +++ b/trunk/lib/lck.c @@ -773,6 +773,11 @@ saLckResourceLock ( goto error_exit; }

[Openais] [lck 4/4] Test lockStatus

--- trunk/lib/lck.c |2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/trunk/lib/lck.c b/trunk/lib/lck.c index 4b453fb..95b2c1e 100644 --- a/trunk/lib/lck.c +++ b/trunk/lib/lck.c @@ -773,7 +773,7 @@ saLckResourceLock ( goto error_exit; } - if

[Openais] [lck 3/3] lib/lck: Unlink calls lckLockIdInstanceFinalize

This prevents previous invalid read in close, because previous code doesn't removes lock from list. --- trunk/lib/lck.c |4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/trunk/lib/lck.c b/trunk/lib/lck.c index 4e962dd..e65c82b 100644 --- a/trunk/lib/lck.c +++

[Openais] [corosync + openais] [patch] - Use only /etc/corosync

, only forward-ported to current trunk. If we want this feature to 1.0 (and don't find any problems) please ACK so I can commit. Regards, Honza From 8e3c925c619265f6cab4cfefe0d64d081716a439 Mon Sep 17 00:00:00 2001 From: Jan Friesse jfrie...@redhat.com Date: Thu, 11 Jun 2009 17:32:52 +0200 Subject

Re: [Openais] [PATCH] autogen.sh: simply to a one-liner: autoreconf -i

Jim, OpenAIS need some love too. Jim Meyering wrote: Jan Friesse mentioned that the existing use of which in autogen.sh was causing trouble on a fresh F11 install (it's not installed). I propose to replace that file with a one-liner: From d0c4fa62fadebaa801997a619e7b2bc2cd8c14d9 Mon Sep

[Openais] [corosync] [patch] - HDB First run

This patch fixes initialization of lock. Currently, lock is initialized only in handle_create. Of course, this is proper way, but some testing programs can use directly hdb_get (before initialization of lock) and then, problem occurs. This patch adds initialization to every functions, which tries

Re: [Openais] call for roadmap features for future releases

Steven Dake wrote: I invite all of our contributors to help define the X.Y roadmap of both corosync and openais. Please submit your ideas on this list. Some examples of suggested ideas have been things like converting to libtool. Also new service engine ideas are highly welcome. Keep ideas

Re: [Openais] [corosync] [patch] - hdb_handle_refcount_get - initialize and lock

Commited with lock only (next patch solves init of lock). Steven Dake wrote: Sorry I missed the hdb_create that is wrong please remove that regards -steve On Wed, 2009-06-24 at 16:00 +0200, Jan Friesse wrote: hdb_handle_refcount_get uses handle without lock and do unlock

Re: [Openais] [corosync] [patch] - HDB First run

Commited with only hdb_database_lock_init call Steven Dake wrote: Honzaf, This patch isn't right. This will destroy the destructor information when calling hdb_create. Just call hdb_lock directly Regards -steve On Wed, 2009-06-24 at 15:59 +0200, Jan Friesse wrote: This patch fixes

[Openais] [corosync] [patch] Deliver messages only from known nodes

Currently, messages are delivered to application without testing, if we know the node (this means, we have no process info about from that node). Following patch fixes that. From 3098d6f74e0fc6ae4f7382f0d7391c461e66fddf Mon Sep 17 00:00:00 2001 From: Jan Friesse jfrie...@redhat.com Date: Thu, 25

Re: [Openais] [corosync] [patch] - Fix problems with long token timeout and cpg

David, More or less agree, but does this patch fixed problem for you or not? David Teigland wrote: On Wed, Jul 01, 2009 at 01:46:03PM -0500, David Teigland wrote: other nodes should immediately recognize it has previously failed and process a complete failure for it. i.e. the full

[Openais] [whitetank] [patch] - Don't segfault when ais group doesn't exists

This patch fixes rhbz#509180 by printing main_config-group rather then group-gr_name (group == NULL) Regards, Honza commit eb69c58dfa9889b7a388728916894fd3465a8024 Author: Jan Friesse jfrie...@redhat.com Date: Tue Jul 7 11:57:47 2009 +0200 Remove segfault when ais group doesn't exists

Re: [Openais] [PATCH]: openais/trunk: Remove resource from cleanup list on close

wrote: Where is the BZ? I have no information about what the problem is, how to reproduce it, etc. I really need to try to fix this today since we are hoping to release today. Please provide information ASAP. On Wed, Jul 08, 2009 at 05:48:21PM +0200, Jan Friesse wrote: Ryan, this patch causes

[Openais] [corosync] [patch] - Allow clean build without error building of stress_cpgcontext

This patch adds -lcoroipcc as build requirement, what is needed in case, when libcoroipcc isn't yet installed on system. Regards, Honza Index: trunk/test/Makefile.am === --- trunk/test/Makefile.am (revision 2359) +++

[Openais] [whitetank] [patch] Backport 2304 to whitetank - detect pause

See SUBJ and Patch Regards, Honza commit ee0023d2592c4f17924a7bb82fd795329b7cc05b Author: Jan Friesse jfrie...@redhat.com Date: Tue Jul 14 11:51:18 2009 +0200 Ability to detect process pause Add ability to detect process pause and not implode the membership algorithm when

[Openais] [corosync] - Cpg synchronization patch

Cpg synchronization patch for conf change messages. The root of the theoretical problem is that cpg_join or cpg_leave messages are being sent via the C apis between synchronization. With the current cpg, synchronization happens in confchg_fn, and then later in cpg_sync_process. cpg_sync_process

[Openais] [whitetank] - cpg synchronization patch

Cpg synchronization backport for whitetank. Regards, Honza commit e7d283f1e968f3c35944d9ebe4a472bface11069 Author: Jan Friesse jfrie...@redhat.com Date: Thu Jul 16 14:08:56 2009 +0200 Backport of send downlist patch diff --git a/branches/whitetank/exec/cpg.c b/branches/whitetank/exec

[Openais] [corosync] - Allow only one connection per (node, pid, grp)

make any sense). Regards, Honza commit b357222ce691be531b8a1ba57ea1ec92898c3536 Author: Jan Friesse jfrie...@redhat.com Date: Thu Jul 16 18:12:21 2009 +0200 Allow only one connection per (node, pid, grp) This patch allows only one connection per (node, pid, grp_name) tuple

[Openais] [openais] Backport of cpgverify to whitetank

This can be really useful for testing. Regards, Honza commit 4adabff870204a7a91a0310805d373511613e120 Author: Jan Friesse jfrie...@redhat.com Date: Mon Jul 20 14:34:25 2009 +0200 Backport of cpgverify to whitetank diff --git a/branches/whitetank/test/Makefile b/branches/whitetank/test

Re: [Openais] [corosync] - Allow only one connection per (node, pid, grp)

Committed revision 2364. Steven Dake wrote: On Mon, 2009-07-20 at 10:03 +0200, Jan Friesse wrote: Patch solves problem, when one process connect multiple times to one group by disallow this situation. Please see patch comment for more informations. David, do you agree, that this is how cpg

[Openais] [corosync] [patch] - Use clock_gettime for timers

if clock_gettime and CLOCK_MONOTONIC exists on system. If not, gettimeofday will be used (so you will get old behavior). Comments are welcomed. Regards, Honza commit 5d2bd8a98f378452a0a7737b7bc990a0c22a2b2f Author: Jan Friesse jfrie...@redhat.com Date: Tue Jul 21 15:33:52 2009 +0200 Support

Re: [Openais] [corosync] [RFE] - added support for service.d

Committed as 2368 Steven Dake wrote: good for commit regards -steve On Tue, 2009-07-21 at 18:06 +0200, Jan Friesse wrote: This should make debianists happy. Regards, Honza ___ Openais mailing list Openais@lists.linux-foundation.org

[Openais] [corosync] [patch] - Handle not entered ver: in service section

SSIA Regards, Honza commit e3e41c39ad213d5487b22fa0fe2d3d25d8096c47 Author: Jan Friesse jfrie...@redhat.com Date: Wed Jul 22 10:33:46 2009 +0200 Handle NULL ver string in service In case user not enter ver: in service section, whole corosync falled with segfaul, because atoi

Re: [Openais] [corosync] [patch] - Use clock_gettime for timers

, doesn't look good for me. 8) after reviewing the changes, I'm not convinced the absolute timers work as expected by the coroapi. good work I think now will. Regards -steve Regards, Honza On Tue, 2009-07-21 at 15:42 +0200, Jan Friesse wrote: Patch should solve problem

Re: [Openais] [corosync] [patch] - Use clock_gettime for timers

Committed revision 2373 Steven Dake napsal(a): Patch looks good for commit Regards -steve On Thu, 2009-07-23 at 10:09 +0200, Jan Friesse wrote: I hope final version of patch. #ifdefs removed and fixed problem with computing HZ (not used, but may be in future) Regards, Honza

[Openais] [corosync] [RFE] - Support for -v (version) option

See SUBJ and Patch. Regards, Honza commit 01cbe593996c01af1c5fa1e507954f27a9abd106 Author: Jan Friesse jfrie...@redhat.com Date: Tue Jul 28 11:56:15 2009 +0200 Added support for -v (version) feature This can be usefull for easier way to get informations of the corosync

Re: [Openais] [corosync] [RFE] - Support for -v (version) option

Fabbio pointed to better solution of this problem. Patch attached. Regards, Honza Jan Friesse wrote: See SUBJ and Patch. Regards, Honza ___ Openais mailing list

Re: [Openais] [openais] [trunk] - ckpt - Compare correct checkpoint_creation_attributes

Committed revision 2047 Steven Dake wrote: good for merge regards -steve On Wed, 2009-07-29 at 10:13 +0200, Jan Friesse wrote: Solves problem with SAF test. Now we are able (sometimes) to make 100%. regards, Honza ___ Openais mailing

Re: [Openais] [corosync] [patch] - Use clock_gettime for timers

. Regards -steve On Wed, 2009-07-29 at 18:33 -0700, Steven Dake wrote: Honza, After more testing, we need the special casing. MacOSX 10.5.7 (latest version) doesn't support clock_gettime. Regards -steve On Mon, 2009-07-27 at 12:13 +0200, Jan Friesse wrote: Committed revision 2373

Re: [Openais] [corosync] [RFE] - Support for -v (version) option

Patch version 3. Uses approach of - delete exec/version.h from svn control - generate it from configure.ac AND svnversion in autogen.sh Regards, Honza Jan Friesse wrote: Steve, I'm looking on tagging script (I hope it's Makefile in root). First thing: - Do we really need patch to show

Re: [Openais] [corosync] [patch] - Use clock_gettime for timers

at 11:51 +0200, Jan Friesse wrote: Steve, what do you think about following patch. No autotools integration, rather one simple #ifdef and warning on compilation. Regards, Honza Steven Dake wrote: After more looking, there is a sysconf test for this (_POSIX_TIMERS). I'm not sure how to do

Re: [Openais] [corosync] [RFE] - Support for -v (version) option

Committed revision 2381. Steven Dake wrote: good for merge regards -steve On Mon, 2009-08-03 at 11:46 +0200, Jan Friesse wrote: Little fix for patch. In case, user rebuild configure.ac and doesn't have svnversion command, display exported. Regards, Honza Jan Friesse napsal

Re: [Openais] [corosync] [patch] - Support for cpg joinlist from/to BE architectures

Committed revision 2384 with comment removed. Regards, Honza Steven Dake napsal(a): agree On Mon, 2009-08-10 at 13:28 +0100, Christine Caulfield wrote: On 10/08/09 13:14, Jan Friesse wrote: See patch. If you're doing that you might as well fix the comment at the same time

[Openais] [whitetank] [patch] - Backport of better limits patch

See patches. I split it to two parts, where: part 1 = Functionality part 2 = cpgbound Regards, Honza From 512a6df157634e7bd0230c8f55965a6bfc9423ad Mon Sep 17 00:00:00 2001 From: Jan Friesse jfrie...@redhat.com Date: Thu, 13 Aug 2009 15:27:38 +0200 Subject: [PATCH 1/2] Better limits for OpenAIS

[Openais] [corosync] [patch] - set found_service_ver to NULL

This patch set found_service_ver to NULL (exec/service.c corosync_service_defaults_link_and_init) so if we don't find ver in objdb, we don't read from uninitialized memory. Regards, Honza commit 37af5c2ad8ead51f8da44cddd37be8d0419aecca Author: Jan Friesse jfrie...@redhat.com Date: Tue Aug 18

[Openais] [corosync] [rfe] - cpg_iteration_* API draft

This feature should allow app to get cpg groups and members. It's modeled like ckpt section iteration, so: Data types: /* Handle to cpg_iteration instance */ typedef uint64_t cpg_iteration_handle_t; /* Type of iteration */ typedef enum { CPG_ITERATION_NAME_ONLY = 1, /* We will return

Re: [Openais] [corosync] [patch] - Fix semaphores on systems without POSIX_THREAD_PROCESS_SHARED

Committed revision 2390. Steven Dake wrote: pls merge regards -steve On Tue, 2009-08-18 at 17:52 +0200, Jan Friesse wrote: This fix one sem_num, so it's now possible to run more then once ipc. Regards, Honza plain text document attachment (coroipcs-system-without-unamed-shared

[Openais] [corosync] [patch?] - Handle NULL callbacks in cpg, confdb, evs services

(user wanted ONE dispatch, not zero), but I'm not sure. Is it possible to make behavior same? Regards, Honza From 25e448fa78560fbdc7be5ad5ac460e8f9939d0ef Mon Sep 17 00:00:00 2001 From: Jan Friesse jfrie...@redhat.com Date: Wed, 19 Aug 2009 15:29:30 +0200 Subject: [PATCH 1/3] cpg_initialize

Re: [Openais] [corosync] [rfe] - cpg_iteration_* API draft

David David Teigland wrote: On Tue, Aug 18, 2009 at 06:09:38PM +0200, Jan Friesse wrote: This feature should allow app to get cpg groups and members. It's modeled like ckpt section iteration, so: Thanks! I've been waiting a long time for this (the same is needed for various other things

[Openais] [corosync] [rfe] - Implementation of cpg_iteration functions

Implementation of DRAFT API. Look like there are no leaks, valgrind problems, ... Regards, Honza From 02a9fbad2db3ee019c735422dc064fed0179e103 Mon Sep 17 00:00:00 2001 From: Jan Friesse jfrie...@redhat.com Date: Wed, 19 Aug 2009 15:17:27 +0200 Subject: [PATCH] Implementation of cpg_iteration

Re: [Openais] [corosync] [patch?] - Handle NULL callbacks in cpg, confdb, evs services

Changed: - break to continue (in cfg too) - (!something) to (something == NULL) Committed as 2392. Steven Dake napsal(a): On Wed, 2009-08-19 at 15:55 +0200, Jan Friesse wrote: Attached patches handle NULL callbacks in [cpg|evs|confdb]_initialize and *_dispatch. Handling is same as in cfg

Re: [Openais] [corosync] [rfe] - Implementation of cpg_iteration functions

void *message) + const void *message should be at 1 tabtop not tabbed to align. It is copying neighboring code. But yes, lib style looks nicer, so changed now. Regards, Honza commit 6e9292f426aaf32da5a12e0f7df9724c2edfd83d Author: Jan Friesse jfrie...@redhat.com Date: Thu Aug 20 15:06

Re: [Openais] [corosync] [rfe] - Implementation of cpg_iteration functions

Fabio, when will libtool happened? I would like to have this feature + tool in Corosync 1.1.0. Is attached patch good solution of problem? Regards, Honza Fabio M. Di Nitto napsal(a): On Wed, 2009-08-19 at 21:57 -0700, Steven Dake wrote: On Thu, 2009-08-20 at 06:39 +0200, Fabio M. Di

  1   2   3   4   >