Hi,
Here's some explain of this:
The 'wsize' and 'rsize' negotiated between client and server, are affected
by the auto-sized memory available to incoming request to NFS server.
And the size is actually adjusted based on total available memory to the server.
The problem is when the client
2008/10/27 Lars Marowsky-Bree [EMAIL PROTECTED]:
On 2008-10-16T17:28:54, Xinwei Hu [EMAIL PROTECTED] wrote:
sfex_daemon.c:288: warning: format '%d' expects type 'int', but argument 5
has type 'long unsigned int'
sfex_daemon.c:357: warning: ignoring return value of 'daemon', declared
2008/10/27 Lars Marowsky-Bree [EMAIL PROTECTED]:
On 2008-10-27T14:24:53, Lars Marowsky-Bree [EMAIL PROTECTED] wrote:
On 2008-10-16T17:28:54, Xinwei Hu [EMAIL PROTECTED] wrote:
sfex_daemon.c:288: warning: format '%d' expects type 'int', but argument 5
has type 'long unsigned int
. ocf_log debug output from init_script
. umount rpc_pipefs when unbinding the tree
2008/10/22 Lars Marowsky-Bree [EMAIL PROTECTED]:
On 2008-10-21T10:56:41, Xinwei Hu [EMAIL PROTECTED] wrote:
Hi Xinwei,
- How is it stacked with IPaddr? It seems to require the IPaddr to be
online before
Hi all,
The attached resource agent is an implementation of shared mode for
storage controlling based on scsi-3 reservation.
The basic idea is to use the notify() operation to handle the
post-stop message, to get a full list of all active nodes, then
preempt in-active-but-still-registered
Hi all,
The resource agent of scsi-2 reservation implements a exclusive mode
for controlling shared storage. It has to be exclusive because it's
the only mode that SCSI-2 reservation supports. There are still a lot
of legacy storages support SCSI-2 only out there, so this resource
agent should
Hi all,
I went through the nfsserver resource agent completely, and here's
the change list
from last time.
. a configurable nfs_init_script replaces the hard coded /etc/init.d/nfsserver
. a configurable nfs_notify_cmd replaces the hard coded /sbin/sm-notify
. nfs_shared_infodir replaces the
2008/10/16 Raoul Bhatia [IPAX] [EMAIL PROTECTED]:
hi,
do you mind me asking what the purpose of sfex is?
sfex implements a advisory protocol over shared disk. It helps to
prevent concurrent accessing to the shared storage even when the
split-site happens.
cheers,
raoul
Xinwei Hu wrote
Hi Lars,
Thanks for the review.
There are several things I still need your help to be made clear.
And about all other bugs not mentioned here, I'll fix and send out a
updated version. :)
2008/10/17 Lars Marowsky-Bree [EMAIL PROTECTED]:
On 2008-10-16T17:28:54, Xinwei Hu [EMAIL PROTECTED
2008/10/6 Lars Marowsky-Bree [EMAIL PROTECTED]:
On 2008-10-06T12:32:11, Yan Gao [EMAIL PROTECTED] wrote:
Hi,
On Sat, 2008-10-04 at 17:39 +0900, [EMAIL PROTECTED] wrote:
Hi Xinwei,
I understood it.
Then I will use latest GUI next week.
If there is a problem in GUI, I report it in a
Hi Hideo,
tip of pygui repository supports the latest version of pacemaker in
the build-service.
It also features some GUI redesigns to improve the usability. ;)
2008/10/3 HIDEO YAMAUCHI [EMAIL PROTECTED]:
Hi Andrew,
Only if you set validate-with=pacemaker-0.6 (which also prevents you
2008/10/2 Dejan Muhamedagic [EMAIL PROTECTED]:
I attached a fixed version for your further comments.
And some are replied below
Thanks for your time to do the review ;)
Relative to where? Why and under which circumstances would one
want to modify this path? Why is it required?
I guess these 2
snmp agent is now separately maintained in pygui.
we are trying to keep snmp_agent update with pacemaker,
comments/suggestions/patches are welcome to snmp agent.
2008/9/29 Michael Schwartzkopff [EMAIL PROTECTED]:
Am Montag, 29. September 2008 12:41 schrieb Lars Marowsky-Bree:
Hi,
in
2008/9/30 Dejan Muhamedagic [EMAIL PROTECTED]:
Hi Xinwei,
On Sun, Sep 28, 2008 at 10:30:07PM +0800, Xinwei Hu wrote:
Hi all,
Sorry that I forget to attach the patch for the previous mail.
The DRAC3 stonith plugin shipped with heartbeat seems not to work with
DRAC5.
The attached
Hi all,
I send an implementation of nfsserver to the user list a while ago.
As there's no interests show around, I resend it here as a patch.
Please consider it for upstream. Thanks.
# HG changeset patch
# User [EMAIL PROTECTED]
# Date 1222583225 -28800
# Node ID
Hi all,
Sorry that I forget to attach the patch for the previous mail.
The DRAC3 stonith plugin shipped with heartbeat seems not to work with DRAC5.
The attached patch is a simple external stonith plugin. It sshes
into the specified IP address, then control the power of the host.
Please
Hi all,
The DRAC3 stonith plugin shipped with heartbeat seems not to work with DRAC5.
The attached patch is a simple external stonith plugin. It sshes
into the specified IP address, then control the power of the host.
Please consider it for upstream.
Hi all,
Following is an explanation of my idea about implement scsi-3
reservation as resource agent. I'd like to present them here, and have
your comments.
===
. basic understanding of scsi-3 reservation type 5:
- hosts can register themselves with their unique host_keys
-
Hi all,
I'm trying to move monitor operation from RA into the application
itself. And, as far as I remember, we can send out failure
notification to CRM directly in current pacemaker. Am I right about
this ? And are there any reference to that ? Either source code or
document will do ;)
2008/6/19 Keisuke MORI [EMAIL PROTECTED]:
Hi,
Xinwei Hu [EMAIL PROTECTED] writes:
I'm the one who opposed sfex in the previous discussion.
My point was simple that:
check-and-reserve on disk is not an atomic CAS operation. and lock
based on that may silently cause data corruption
2008/6/19 Dejan Muhamedagic [EMAIL PROTECTED]:
Hi,
On Thu, Jun 19, 2008 at 09:26:13PM +0800, Xinwei Hu wrote:
2008/6/19 Keisuke MORI [EMAIL PROTECTED]:
Hi,
Xinwei Hu [EMAIL PROTECTED] writes:
I'm the one who opposed sfex in the previous discussion.
My point was simple
2008/6/20 Lars Marowsky-Bree [EMAIL PROTECTED]:
On 2008-06-19T22:52:55, Xinwei Hu [EMAIL PROTECTED] wrote:
True. It is possible to break sfex, but the probability that that
is going to happen is extremely low and could be due only to a
very pathological timing. One way to make
I'm the one who opposed sfex in the previous discussion.
My point was simple that:
check-and-reserve on disk is not an atomic CAS operation. and lock
based on that may silently cause data corruption.
I haven't follow the evolution of sfex though, so things might have
been changed.
Just FYI.
refer to
rgmanager/src/resource/nfsserver.sh
commit 8a8b5bb539c09a2c504c2ceb81548f508d5f11c0
2008/6/3 Serge Dubrouski [EMAIL PROTECTED]:
On Tue, Jun 3, 2008 at 3:25 AM, Xinwei Hu [EMAIL PROTECTED] wrote:
clurmtabd is deprecated by rgmanager itself already (refer to git
version of rgmanager
to rpc.statd.
On Tue, Jun 3, 2008 at 9:27 AM, Serge Dubrouski [EMAIL PROTECTED] wrote:
On Tue, Jun 3, 2008 at 8:52 AM, Xinwei Hu [EMAIL PROTECTED] wrote:
refer to
rgmanager/src/resource/nfsserver.sh
commit 8a8b5bb539c09a2c504c2ceb81548f508d5f11c0
Do you have a link for it? I can't find
2008/6/4 Serge Dubrouski [EMAIL PROTECTED]:
On Tue, Jun 3, 2008 at 9:53 AM, Xinwei Hu [EMAIL PROTECTED] wrote:
2008/6/3 Serge Dubrouski [EMAIL PROTECTED]:
I found it, though it was a commit for nfsexport.sh. One thing isn't
clear though.
Now they prefer to bind /var/lib/nfs to a subdir under
Hi all,
I'm drafting a document on parameters passed to RA. Besides those mentioned
on opencf.org and linux-ha.org, I found a lot of CRM_meta_xxx parameters which
don't explained anyway (or I just missed ;-/)
Here's a list of what I find out. Can someone explain it a bit ?
Hi all,
The ibmhmc stonith plugin doesn't work with the web interface of HMC.
The attachment is a workable version of stonith plugin through HMC
web interface. It depends on curl and /bin/sh.
It'll be great if someone can help to review and include it upstream then.
Thanks.
hmchttp
2007/12/6, Andrew Beekhof [EMAIL PROTECTED]:
done:
http://hg.beekhof.net/lha/crm-dev/rev/6e4c6fb9d160
Thanks :)
However, I don't think reusing '--force-quorum ' is a good approach,
that'll create
confusions. Can you make some change to that ?
Regards,
Lars
--
Teamlead
2007/12/4, Andrew Beekhof [EMAIL PROTECTED]:
On Dec 3, 2007, at 7:23 PM, Dejan Muhamedagic wrote:
Hi,
On Tue, Dec 04, 2007 at 12:20:15AM +0800, Xinwei Hu wrote:
Hi all,
We have a instance about cibadmin recently. A typo of 'cibadmin -r
blahblah' forces the HA into RO mode
/
nvpair name=target_role
id=stonith_APS06:0_target_role value=started/
/attributes
/instance_attributes
/primitive
/clone
2007/11/8, Andrew Beekhof [EMAIL PROTECTED]:
On Nov 8, 2007, at 9:51 AM, Xinwei Hu wrote:
Hi all,
While setting
Hi all,
While setting no_quorum_policy to freeze stonith_enabled,
heartbeat keeps complaining:
ERROR: unpack_operation: Stonith resources (eg. stonith_APS06:3)
cannot require fencing to start
It's caused by a logic error(to me at least), which can be fixed by
the attached patch.
Hi Andrew,
So can you help to review the patch or give a better solution for that ?
Thanks.
2007/11/8, Andrew Beekhof [EMAIL PROTECTED]:
On Nov 8, 2007, at 11:03 AM, Xinwei Hu wrote:
Hi Andrew,
The only resources are clone sets of riloe stonith:
clone ordered=true id
Alan,
2007/8/17, Alan Robertson [EMAIL PROTECTED]:
Xinwei Hu wrote:
You can also do a ucast 127.0.0.1 without any new code... That's what
we do for testing in BasicSanityCheck.
Thanks for the tip.
--
Alan Robertson [EMAIL PROTECTED]
Openness is the foundation and preservative
2007/8/13, Junko IKEDA [EMAIL PROTECTED]:
Assume we have 2 nodes.
1. Node A B reach step 3) in the same time.
2. sfex_lock on Node B is scheduled out due to some other reasons.
3. sfex_lock on Node A goes through step 3 to 6, and Node A holds
the lock now.
Node A is sure to hold
Hi all,
I'm trying to run haclient on the Windows platform. The biggest difficulty
is to port the C code of lib/mgmt to mingw/cygwin platform.
A much eaiser step, which I take now, is to rewrite the client part of
lib/mgmt in pure python (ctypes + gnutls binding). The windows version of
On 2007-06-12T03:33:33, Xinwei Hu [EMAIL PROTECTED] wrote:
Hi all,
I'm trying to run haclient on the Windows platform. The biggest
difficulty is to port the C code of lib/mgmt to mingw/cygwin platform.
A much eaiser step, which I take now, is to rewrite the client part of
lib
Hi all,
Here's the updated version. It can work on a 2-vmware-nodes hb2 cluster
with
several dummy resources running for hours now. :)
2007/6/1, Lars Marowsky-Bree [EMAIL PROTECTED]:
On 2007-06-01T17:22:24, Xinwei Hu [EMAIL PROTECTED] wrote:
Random comments, which I've not sorted
pkts missing.
Such a simple protocol is used because I failed to google out something
better. :(
You suggestions on how to improve are really appreciated. Thanks.
/*
* dskcm.c: Disk based communication code for heartbeat.
*
* Copyright (C) 2007 Xinwei Hu [EMAIL PROTECTED
39 matches
Mail list logo