ci-linux-devel — Discussion about Cluster Infrastructure for Linux

[CI] Commercial Interest in ci linux

[Rick B. <ri...@ra...>]

Hello, are there any contractors/freelancers/consultants who work with
this project and would be interested in talking to me about its
capabilities and stability? I am happy to pay you for your time.

http://ci-linux.sourceforge.net/





-----------------
Thank-You
Rick Blundell
RapidVPS.com
ri...@ra...
+1 407 442 2898

[CI] [ ci-linux-Feature Requests-2881140 ] Support TCP sendpage

[SourceForge.net <no...@so...>]

Feature Requests item #2881140, was opened at 2009-10-17 22:04
Message generated for change (Comment added) made by rogertsang
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=405833&aid=2881140&group_id=32543

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Priority: 5
Private: No
Submitted By: Roger Tsang (rogertsang)
Assigned to: Roger Tsang (rogertsang)
Summary: Support TCP sendpage

Initial Comment:
Add ability to transport vector of pages directly on the wire via native TCP/IP sendpage()

----------------------------------------------------------------------

>Comment By: Roger Tsang (rogertsang)
Date: 2009-10-27 00:00

Message:
checked-in

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=405833&aid=2881140&group_id=32543

[CI] [ ci-linux-Feature Requests-2881140 ] Support TCP sendpage

[SourceForge.net <no...@so...>]

Feature Requests item #2881140, was opened at 2009-10-17 22:04
Message generated for change (Tracker Item Submitted) made by rogertsang
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=405833&aid=2881140&group_id=32543

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Priority: 5
Private: No
Submitted By: Roger Tsang (rogertsang)
Assigned to: Roger Tsang (rogertsang)
Summary: Support TCP sendpage

Initial Comment:
Add ability to transport vector of pages directly on the wire via native TCP/IP sendpage()

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=405833&aid=2881140&group_id=32543

[CI] WAN usage posible?

[Digital W. <dig...@ho...>]

 

Hi there:

 

I am new to OpenSSI and CI so please forgive me if my questions are obvious, or just "stupid".

I am reading everything I can find about it but didn't find a definitive answer for it...

 

The question is:
Can OpenSSI / CI be used to leverage servers over a WAN?

The objective would be to load balance webservers / APPservers with session continuity distributing jobs to the most apropiate server based on loads, network saturation, geolocation or whatever helps the webpage to arrive at the client browser faster.


Thank you.




 


--


Yosu Cadilla
605 304 382
MSN: dig...@ho...
MSN: yos...@ho...
AIM: BizBoost
Yahoo: bizboost
ICQ: 438468115

 



BizBoost S.L.
http://www.bizboost.net

Avenida del Cid, 17
46018 VALENCIA
963 822 646






_________________________________________________________________
Show them the way! Add maps and directions to your party invites. 
http://www.microsoft.com/windows/windowslive/products/events.aspx

Re: [CI] hello ?

[Greg F. <gre...@gm...>]

2008/1/12 Djuro Kolundzija <dju...@gm...>:
> HI,
>
>
> I have a question about linux cl. I wont to setup web server but I need
> about 10 TR of space, now  1 machine cant hold  that many  HD, as fur as I
> know, now, can linux cl  connect more  than  one machine (which it does)
> only do they all show as 1 big hard disk when  I use it as web server.?  I
> hope you understand my question, because I need lot of space does the linux
> cl also put all the space together so I can use as web server, or  it only
> helps with work  load ??  Please let me know  if you can  anything.   Thank
> You So Much in Advance !!! Have A Good Day !

OT:
I assume you meant 10 TB of disk space.  Why is that a unsolvable
problem in one machine?  Seems much easier than the above.

Even with 500GB drives, you could do 10TB with 24 drives in some
reasonably safe form of raid.

Having 24 drives is not that hard.  You can buy a big rack mount
chassis with 24 disk slots and put your own MB in it, or you can get
multiple PMP enclosures and connect them to your box by eSata.

PMP is supported in 2.6.23 kernels, or is supported in OpenSUSE 10.3
with a patched 2.6.22 kernel.  Nothing older.  Check the linux-ide
kernel archives to see what PMP enclosures have been tested.  Tejun
Heo is the main expert on the subject and wrote the kernel support.

Greg
-- 
Greg Freemyer
Litigation Triage Solutions Specialist
http://www.linkedin.com/in/gregfreemyer
First 99 Days Litigation White Paper -
http://www.norcrossgroup.com/forms/whitepapers/99%20Days%20whitepaper.pdf

The Norcross Group
The Intersection of Evidence & Technology
http://www.norcrossgroup.com

[CI] hello ?

[Djuro K. <dju...@gm...>]

HI,


I have a question about linux cl. I wont to setup web server but I need
about 10 TR of space, now  1 machine cant hold  that many  HD, as fur as I
know, now, can linux cl  connect more  than  one machine (which it does)
only do they all show as 1 big hard disk when  I use it as web server.?  I
hope you understand my question, because I need lot of space does the linux
cl also put all the space together so I can use as web server, or  it only
helps with work  load ??  Please let me know  if you can  anything.   Thank
You So Much in Advance !!! Have A Good Day !

[CI] CLMS stuck in pre-root initialization on 2-node SSI cluster

[Roger T. <rog...@gm...>]

Hi,

I ran into the following problem on a 2-node DRBD-SSI cluster.  The
clms master was failing over, but paniced due to IO errors.  At the
same time the client was booting but did not detect the root node went
down.  Last console message was waiting to join cluster.  Not good,
but hard to reproduce.

The clms master must have failed before nodedown daemon spawned or the
daemon never got the event.  So the client got stuck in an infinite
loop somewhere I think in clms_client_sync_masterparams().

Roger


--- cluster/clms/clms_client.c  10 Feb 2005 01:05:32 -0000      1.7
+++ cluster/clms/clms_client.c  18 Feb 2007 06:57:02 -0000
@@ -158,7 +158,7 @@
 static void
 clms_client_sync_masterparams(void)
 {
-       int error;
+       int error, tries = 0;
        int rval;
        char *masterinfo;
        int masterinfo_len;
@@ -185,11 +185,17 @@
                }
                if (error == -EREMOTE) {
                        nidelay(HZ);
+                       if (++tries > CONFIG_NODE_MONITOR_TIMEOUT_MS /
1000 + 1) {
+                               printk(KERN_WARNING
+                                      "Failed reading master list,
master went down!\n");
+                               machine_restart(NULL);
+                       }
                        continue;
                }
                if (error >= 0)
                        error = rval;
                if (error < 0)
+                       /* XXX: Never reached */
                        panic("%s:Error %d reading master list\n",
                              __FUNCTION__, -error);
                break;

[CI] HELLO I have important question.

[tomekbu <to...@dg...>]

Hello!!

I am obligate to create hot standby on linux for my two servers. 
Everytying would be ok if i wouldnt use database there. I found OpenSSi 
and it reallly solve my problem (cluster would be great solution). But i 
have question.
There are two nodes in cluster. 1 and 2.
1 node mounts database dictionary (POSTGRESQL) in cluster, so now 1st 
and 2nd can use it and make changes. Now let say that 1st node failed. 
If Node nr 2 would sitll be able to use whole database??

PLEASE help me, its very improatnt ...


Best  Regards,
Tomasz Burda

[CI] Infiniband updates for ICS at git.openssi.org

[Aneesh K. <ane...@gm...>]

Hi Stan,

Today i merged the chnages that you send me across. Since the code
base i have is somewhat different from the ICS in the CVS can you
verify if i get all the changes correct.


commit:

http://git.openssi.org/~kvaneesh/gitweb.cgi?h=ee9b4c733a0c781e457a77bdc8a95c789a49f918;p=ci-to-linus.git;a=commit


diff:
http://git.openssi.org/~kvaneesh/gitweb.cgi?h=ee9b4c733a0c781e457a77bdc8a95c789a49f918;p=ci-to-linus.git;hp=f880e30e944f4c857b775ec531262fa5bd67258d;a=commitdiff

I have done a compile test with the changes. I will give it a try once
i get access to the cluster having infiniband interconnect.

-aneesh

Re: [CI] [Announce] Linux kernel Cluster Framework 0.2

[Aneesh K. <ane...@gm...>]

On 9/16/06, Aneesh Kumar <ane...@gm...> wrote:
> This is the 0.2 release of Linux kernel Cluster Framework.
>
> What is LKCF:
> -----------
> LKCF's aim is to provide a transport independent cluster communication
> framework within the kernel. This enables the developers to write
> kernel based cluster services without being worried about
> communication transport. It also support RPC style programming. That
> means to write kernel service one need to write the service definition
> file (<service>.svc>) and the implementation API. LKCF framework will
> generate all the  registration routines and the marshaling code. Also
> it takes care of forwarding the SIGNALs across different nodes.The
> particular service can be called from any node specifying the node at
> which this particular service need to be executed. All the underlying
> management interface is taken care by the LKCF.
>

Attaching the short log below

commit 5cfb96a487d1ca1893c30ca112706a315c492197
Author: Aneesh Kumar K.V <ane...@gm...>

    A simple test case for ICS.

commit 8ebdda8eca87867b1b6123ab08e1e270f419c74a
Author: Aneesh Kumar K.V <ane...@gm...>

    Token facility needed for cluster based synchronization

commit c7214483043dbc1fce194d0ead4be7d30d587dcf
Author: Stan Smith <sta...@in...>

    ICS over Infiniband verbs work-in-progress.

commit b35812c68c33357335d9acd44ce702f27c87bbc3
Author: Aneesh Kumar K.V <ane...@gm...>

    Internode communication subsystem for Linux.

-aneesh

Re: [CI] Linux kernel Cluster Framework 0.2

[Aneesh K. <ane...@gm...>]

On 9/16/06, Smith, Stan <sta...@in...> wrote:
>
> This juncture seems like a reasonable time to start a discussion of
> heterogeneous ICS networks beneath the transport independent LCF.
>
> What are your intentions towards supporting heterogeneous ICS networks?
>
> ICS message routing & forwarding?
>
> Thoughts on the subjects?

Linux kernel currently lacks a framework which one can use to write
cluster services. The idea of cleaning up CI is to make sure that we
can use CI as a framework. For that i removed the CLMS from the kernel
and introduced configfs based configuration. The first release which
was released as CI 0.1 contain the result of that work. This confused
with previous releases of CI. So  i decided to rename CI to LKCF. The
name cluster framwork should get some interest in the code too. The
purpose of 0.2 release of LKCF was to share your work of using
infiniband as the interconnect. That goes with the idea of release
early . My idea with LKCF is to take CFS and clusterwide shared memory
to the generic kernel.

-aneesh

[CI] Linux kernel Cluster Framework 0.2

[Smith, S. <sta...@in...>]

This juncture seems like a reasonable time to start a discussion of
heterogeneous ICS networks beneath the transport independent LCF.

What are your intentions towards supporting heterogeneous ICS networks?

ICS message routing & forwarding?

Thoughts on the subjects?

[CI] [Announce] Linux kernel Cluster Framework 0.2

[Aneesh K. <ane...@gm...>]

This is the 0.2 release of Linux kernel Cluster Framework.

What is LKCF:
-----------
LKCF's aim is to provide a transport independent cluster communication
framework within the kernel. This enables the developers to write
kernel based cluster services without being worried about
communication transport. It also support RPC style programming. That
means to write kernel service one need to write the service definition
file (<service>.svc>) and the implementation API. LKCF framework will
generate all the  registration routines and the marshaling code. Also
it takes care of forwarding the SIGNALs across different nodes.The
particular service can be called from any node specifying the node at
which this particular service need to be executed. All the underlying
management interface is taken care by the LKCF.

What transport are supported as of today:
---------------------------------------
IPV4
Infiniband verbs/RDMA.

Project Documentation:
---------------------
http://ci-linux.sourceforge.net/

Project git repository:
---------------------------
http://git.openssi.org/~kvaneesh/gitweb.cgi?p=ci-to-linus.git;a=summary

The patches can be found at
http://git.openssi.org/~kvaneesh/ics_patches/lkcf-0.2/

and include

0001-Internode-communication-subsystem-for-Linux.txt
0002-ICS-over-Infiniband-verbs-work-in-progress.txt
0003-Token-facility-needed-for-cluster-based-synchronization.txt
0004-A-simple-test-case-for-ICS.txt

The important changes with the release is the support of infiniband
interconnect. The changes are originally from stan smith with further
cleanup and fixes from me.

The patches are on top of git SHA1
ef7d1b244fa6c94fb76d5f787b8629df64ea4046 of the linus tree.


-aneesh

[CI] Linux Cluster Framework (LCF) is now ready with two transport.

[Aneesh K. <ane...@gm...>]

Hi All,

Yes it is CI. But i have changed it a lot and that also means CI
services written won't compile with LCF.

The final patches are at.
http://git.openssi.org/~kvaneesh/gitweb.cgi?p=ci-to-linus.git;a=summary

-aneesh

[CI] ICS release [ ics 0.1 ]

[Aneesh K. <ane...@gm...>]

Hi All,

This is the first release of ICS. The related patch files can be found at

http://git.openssi.org/~kvaneesh/ics_patches/ics-0.1/

These patches are against the latest linus tree 2.6.18-rc4. If you
want to get the full kernel you can use the snapshot link at

http://git.openssi.org/~kvaneesh/gitweb.cgi?p=ci-to-linus.git;a=summary

There is no kernel based CLMS here. The membership need to be run in
user space. ICS now uses configfs for cluster configuration.

Simple steps to configure the cluster involve

$more cluster_start
ifconfig eth0 192.168.2.15 netmask 255.255.255.0
mount -t configfs none /config
mkdir  /config/ics/node2
echo -n "192.168.2.15" > /config/ics/node2/icsinfo
echo -n "2" > /config/ics/node2/nodenum
ln -s /config/ics/node2 /config/ics/this_node
echo -n "NODE_PREROOT"  > /config/ics/node2/status
echo -n "NODE_POSTROOT"  > /config/ics/node2/status
echo -n "NODE_UP" > /config/ics/node2/status
echo -n "NODE_READY" > /config/ics/node2/status


To indicate the details about other node

$ more start_othernode
mkdir /config/ics/node3
echo -n "192.168.2.16" > /config/ics/node3/icsinfo
echo -n "3" > /config/ics/node3/nodenum
echo -n "NODE_CONFIG" > /config/ics/node3/status
echo -n "NODE_UP" > /config/ics/node3/status
echo -n "NODE_READY" > /config/ics/node3/status

Please feel free to integrate any other membership service and send
patches across to me. I will add it in the next release.

If you are working on another transport for ICS i would encourage you
to use the above patches so that we have a clean interface to submit
upstream.

The next steps is the CFS, so if you are willing to help let me know.

Also in case you get to write some test case using the above patch
send it across to me. I will add it to the repository.


NOTE: The master branch shown in gitweb doesn't show the full history
of changes you can browse to *-ci-to-linus branches to find the full
history.

-aneesh

Re: [CI] only RedHat packages?

[Aneesh K. <ane...@gm...>]

On 8/24/06, Greg Freemyer <gre...@gm...> wrote:
> Has there been an effort to get CI accepted into mainline?
>
> Is it likely to happen soon?
>

I have cleaned up the code and i belive from the code point of view we
are very much in place . But i guess we need to address the below
before we can go further.

a) No in kernel user.
b) Handle memory allocation failures if possible. Otherwise document
why we can't handle.

-aneesh

Re: [CI] only RedHat packages?

[Greg F. <gre...@gm...>]

Has there been an effort to get CI accepted into mainline?

Is it likely to happen soon?

Greg

On 8/20/06, Aneesh Kumar <ane...@gm...> wrote:
> On 8/20/06, niam. tni <nia...@go...> wrote:
> > Hi!
> >
> > What's about support for Debian Linux?
> > Or is there any generic installer script for Linux?
> >
> > CVS access does not work:
> > cvs -d:pserver:ano...@cv...:/cvsroot/ci-linux login
> > Logging in to :pserver:ano...@cv...:2401/cvsroot/ci-linux
> > CVS password:
> > cvs login: authorization failed: server cvs.openssi.org rejected
> > access to /cvsroot/ci-linux for user anonymous
>
>
> If you are looking for a CI functionality only i would request you to look at
>
> http://git.openssi.org/~kvaneesh/gitweb.cgi?p=ci-to-linus.git;a=summary
>
> I have made quiet a lot of changes. There is no CLMS in the kernel any
> more you should be possible to plug it into anything like heatbeat or
> CMAN from user space.
>
> -aneesh
>
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> ci-linux-devel mailing list
> ci-...@li...
> https://lists.sourceforge.net/lists/listinfo/ci-linux-devel
>


-- 
Greg Freemyer
The Norcross Group
Forensics for the 21st Century

Re: [CI] only RedHat packages?

[Aneesh K. <ane...@gm...>]

On 8/20/06, niam. tni <nia...@go...> wrote:
> Hi!
>
> What's about support for Debian Linux?
> Or is there any generic installer script for Linux?
>
> CVS access does not work:
> cvs -d:pserver:ano...@cv...:/cvsroot/ci-linux login
> Logging in to :pserver:ano...@cv...:2401/cvsroot/ci-linux
> CVS password:
> cvs login: authorization failed: server cvs.openssi.org rejected
> access to /cvsroot/ci-linux for user anonymous


If you are looking for a CI functionality only i would request you to look at

http://git.openssi.org/~kvaneesh/gitweb.cgi?p=ci-to-linus.git;a=summary

I have made quiet a lot of changes. There is no CLMS in the kernel any
more you should be possible to plug it into anything like heatbeat or
CMAN from user space.

-aneesh

[CI] only RedHat packages?

[niam.tni <nia...@go...>]

Hi!

What's about support for Debian Linux?
Or is there any generic installer script for Linux?

CVS access does not work:
cvs -d:pserver:ano...@cv...:/cvsroot/ci-linux login
Logging in to :pserver:ano...@cv...:2401/cvsroot/ci-linux
CVS password:
cvs login: authorization failed: server cvs.openssi.org rejected
access to /cvsroot/ci-linux for user anonymous

regards

[CI] ics configuration via configs.

[Aneesh K. <ane...@gm...>]

With the ics cleanup codebase i now have ics configuration via configfs.

We are now one step close to use a user space membership service.

http://git.openssi.org/~kvaneesh/gitweb.cgi?p=ci-to-linus.git;a=commitdiff;h=f9e64bda84b1cfffe34ecb93fd8ce6d50b41f59f

-aneesh

[CI] Reworked CI code

[Aneesh K. <ane...@gm...>]

Hi,

I  was able to run a simple case with the reworked CI code base. I had
a two node cluster
and then the simple test case that did a remote prink.

I guess now it is time to look at  integrating it with CMAN.

You can find the work at

http://git.openssi.org/~kvaneesh/gitweb.cgi?p=ci-to-linus.git;a=summary


-aneesh

[CI] CI code base

[Aneesh K. <ane...@gm...>]

If you are interested in finding out the details of the cleanup i can
merged the cleanups to 2.6.17-rc4. You can find the diff at


http://git.openssi.org/~kvaneesh/gitweb.cgi?p=3Dci-to-linus.git;a=3Dcommitd=
iff;h=3D5a3c35f45a6f9a781cf26c88abcdaf35d90bd5d8;hp=3D387e2b0439026aa738a9e=
dca15a57e5c0bcb4dfc

Let me know if you have any suggestion.

-aneesh

[CI] Re: dropping XDR from ICS param attributes

[John B. <joh...@hp...>]

Aneesh Kumar wrote:
> On 5/11/06, Aneesh Kumar <ane...@gm...> wrote:
>> On 5/11/06, Aneesh Kumar <ane...@gm...> wrote:
>> > Hi John and Brian,
>> >
>> > What is the specific requirment of using XDR within ICS. ? I see
>> > structures like pt_regs being encoded using ics directly. As per our
>> > last discussion most of these structures are marked OOL. So with XDR
>> > we encode them with XDR routines and out then as OOL data. But then if
>> > all the nodes involved in clustering are of same endian do we need to
>> > encode them using XDR routines ?
>> >
>>
>> After looking at the code again i guess i understand what is
>> happening. For structures carrying pointers to another data type we
>> need the XDR param type.
>>
>> I have dropped from the latest ICS code the openssirpcgen. I am right
>> now looking at NFS and understanding how to do the work the same way
>> they are doing.
>>
>>
> 
> How about doing it the below way
> 
> struct my_struct {
>   int a;
>   char *data;
> }
> 
> cli_encode_xdr_struct_my_struct(cli_handle_t *handle, struct my_struct *p)
> {
>    /* scalar type encode inline */
>    cli_encode_int(handle, p->a)
>    cli_encode_ool_data_t(handle, p->data, length_of_data, NULL,  0);
> }
> 
> 
> 
> By carefully doing this way i guess we should be able to drop XDR and 
> rpcgen

You certainly can. It can become painful for complex structures. You 
don't necessarily have to encode the pointer data as ool, though. It 
depends on the length. (There is a 300 byte limit on inline data.)

John


> 
> 
> -aneesh
> 

[CI] Re: dropping XDR from ICS param attributes

[John B. <joh...@hp...>]

Aneesh Kumar wrote:
> On 5/11/06, Aneesh Kumar <ane...@gm...> wrote:
>> Hi John and Brian,
>>
>> What is the specific requirment of using XDR within ICS. ? I see
>> structures like pt_regs being encoded using ics directly. As per our
>> last discussion most of these structures are marked OOL. So with XDR
>> we encode them with XDR routines and out then as OOL data. But then if
>> all the nodes involved in clustering are of same endian do we need to
>> encode them using XDR routines ?
>>
> 
> After looking at the code again i guess i understand what is
> happening. For structures carrying pointers to another data type we
> need the XDR param type.

You don't need it. XDR is just a convenient way of generating the glue 
for a complex structure. You can always write the marshaling routines by 
hand and drop the XDR.

> 
> I have dropped from the latest ICS code the openssirpcgen. I am right
> now looking at NFS and understanding how to do the work the same way
> they are doing.
> 
> 
> -aneesh
> 

[CI] Re: dropping XDR from ICS param attributes

[Aneesh K. <ane...@gm...>]

On 5/11/06, Aneesh Kumar <ane...@gm...> wrote:
> On 5/11/06, Aneesh Kumar <ane...@gm...> wrote:
> > Hi John and Brian,
> >
> > What is the specific requirment of using XDR within ICS. ? I see
> > structures like pt_regs being encoded using ics directly. As per our
> > last discussion most of these structures are marked OOL. So with XDR
> > we encode them with XDR routines and out then as OOL data. But then if
> > all the nodes involved in clustering are of same endian do we need to
> > encode them using XDR routines ?
> >
>
> After looking at the code again i guess i understand what is
> happening. For structures carrying pointers to another data type we
> need the XDR param type.
>
> I have dropped from the latest ICS code the openssirpcgen. I am right
> now looking at NFS and understanding how to do the work the same way
> they are doing.
>
>

How about doing it the below way

struct my_struct {
   int a;
   char *data;
}

cli_encode_xdr_struct_my_struct(cli_handle_t *handle, struct my_struct *p)
{
    /* scalar type encode inline */
    cli_encode_int(handle, p->a)
    cli_encode_ool_data_t(handle, p->data, length_of_data, NULL,  0);
}



By carefully doing this way i guess we should be able to drop XDR and rpcge=
n


-aneesh