Re: [OpenAFS] forcing and benchmarking the cache scan

2011-03-18 Thread Jason Edgecombe 
Thanks. I got things squared away. I ran "mke2fs -j" once on my cache 
partition to make the scan faster on boot.



On 03/18/2011 10:44 AM, Derrick Brashear wrote:

/etc/init.d/openafs-client stop; rm /usr/vice/etc/CacheItems;
/etc/init.d/openafs-client start
?


On Fri, Mar 18, 2011 at 10:05 AM, Edgecombe, Jasonwrote:


Hi everyone,



I had a server doing a cache scan for over an hour yesterday after
upgrading the RAID (mpt2sas) driver on RHEL5.5 64bit. I thought the box
hung, so I ran a hard reset and ended up rebuilding it. I’m trying to
reproduce and measure the cache scan time with the two different versions of
the RAID driver to see if the driver causes an issue, or not. How can I
force the client to run a cache scan when I run /etc/init.d/openafs-client
restart instead of at boot time?



I’m trying the following procedure to benchmark things:



fs setcachesize 1

fs setcachesize 62048980

fs getcachep

cd /afs/(some folder)

find . -type f | xargs cat>  /dev/null

cd /

time /etc/init.d/openafs-client restart



Details on production box:

OS: RedHat Enterprise Linux Server 5.5 (64bit)

Hardware: Dell PowerEdge R210 With H200 RAID controller

Driver: mp2tsas driver from Dell



Details on testing box (my only H200 cards are in production servers):

OS: RedHat Enterprise Linux Server 5.5 (64bit)

Hardware: Dell PowerEdge R510 With H700 RAID controller

Driver: mp2tsas driver from Dell



Thanks,

Jason



---

Jason Edgecombe *| *Linux and Solaris Administrator

UNC Charlotte *| *The William States Lee College of Engineering

9201 University City Blvd. *| *Charlotte, NC 28223-0001

jwedg...@uncc.edu *| *http://coe.uncc.edu | [image: 
facebook-logo]
  Facebook

---

If you are not the intended recipient of this transmission or a person
responsible for delivering it to the intended recipient, any disclosure,
copying, distribution, or other use of any of the information in this
transmission is strictly prohibited. If you have received this transmission
in error, please notify me immediately by reply e-mail or by telephone at
704-687-3514.  Thank you.








___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info

Re: [OpenAFS] Re: "Failed to get heavyweight reference" when releasing

2011-03-18 Thread Derrick Brashear 
hitting bugs is good!

Derrick


On Mar 18, 2011, at 8:33 PM, "Ryan C. Underwood"  
wrote:

> On Fri, Mar 18, 2011 at 04:28:18PM -0500, Andrew Deason wrote:
>> 
>> Sorry, this is an issue that was fixed on master and we missed the 1.6
>> pullup. This patch should fix it:
>> 
> 
> Yes, that seems to have fixed it!  I seem to be having all the luck
> lately hitting obscure bugs :-)
> 
> -- 
> Ryan C. Underwood, 
> ___
> OpenAFS-info mailing list
> OpenAFS-info@openafs.org
> https://lists.openafs.org/mailman/listinfo/openafs-info
___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info

Re: [OpenAFS] Re: "Failed to get heavyweight reference" when releasing

2011-03-18 Thread Ryan C. Underwood 
On Fri, Mar 18, 2011 at 04:28:18PM -0500, Andrew Deason wrote:
> 
> Sorry, this is an issue that was fixed on master and we missed the 1.6
> pullup. This patch should fix it:
> 

Yes, that seems to have fixed it!  I seem to be having all the luck
lately hitting obscure bugs :-)

-- 
Ryan C. Underwood, 
___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info

[OpenAFS] Multiple logins

2011-03-18 Thread Jaap Winius 

Hi folks,

My site uses OpenAFS and MIT Kerberos with OpenLDAP for user meta data  
(all running on Debian squeeze). Is it possible to prevent users from  
logging in more than once, or at least to prevent them from starting  
up the same desktop environment on multiple hosts with the same  
account, since this usually leads to problems?


Thanks,

Jaap
___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info

[OpenAFS] Re: "Failed to get heavyweight reference" when releasing

2011-03-18 Thread Andrew Deason 
On Fri, 18 Mar 2011 16:10:36 -0500
"Ryan C. Underwood"  wrote:

> 
> Never seen this before.  Related to dafileserver?  I'm getting this in
> FileLog after a server crash and restart:
> 
> Fri Mar 18 16:00:01 2011 FSYNC_com_VolOff: failed to get heavyweight 
> reference to volume 536871132 (state=20, flags=0x18)

Sorry, this is an issue that was fixed on master and we missed the 1.6
pullup. This patch should fix it:


-- 
Andrew Deason
adea...@sinenomine.net

___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info

[OpenAFS] "Failed to get heavyweight reference" when releasing

2011-03-18 Thread Ryan C. Underwood 

Never seen this before.  Related to dafileserver?  I'm getting this in
FileLog after a server crash and restart:

Fri Mar 18 16:00:01 2011 FSYNC_com_VolOff: failed to get heavyweight reference 
to volume 536871132 (state=20, flags=0x18)
Fri Mar 18 16:02:00 2011 FSYNC_com_VolOff: failed to get heavyweight reference 
to volume 536871132 (state=20, flags=0x18)
Fri Mar 18 16:02:02 2011 FSYNC_com_VolOff: failed to get heavyweight reference 
to volume 536871132 (state=20, flags=0x18)
Fri Mar 18 16:03:48 2011 FSYNC_com_VolOff: failed to get heavyweight reference 
to volume 536871132 (state=20, flags=0x18)


At first I was getting this:

# vos exa root_dist
root_dist 536871131 RW 603230 K  On-line
valhalla.icequake.net /vicepa
RWrite  536871131 ROnly  536871132 Backup  0
MaxQuota  0 K
CreationSat Oct 25 22:56:52 2003
CopySun Apr 19 23:12:16 2009
Backup  Tue Jan  6 06:43:01 2004
Last Access Fri Mar 18 15:43:01 2011
Last Update Fri Mar 18 15:00:04 2011
1801 accesses in the past day (i.e., vnode references)

RWrite: 536871131 ROnly: 536871132
number of sites -> 3
   server valhalla.icequake.net partition /vicepa RW Site
   server xanadu.icequake.net partition /vicepa RO Site
   server valhalla.icequake.net partition /vicepa RO Site

# ls /afs/icequake.net/dist
arch  cpan  fonts  pear  src  trees  <-- RO is online

# ls /afs/.icequake.net/dist
arch  cpan  fonts  pear  src  trees  <-- RW is online

# vos release root_dist
Failed to reclone the RW volume 536871131
Volume not attached, does not exist, or not on line
Error in vos release command.
Volume not attached, does not exist, or not on line



Then I removed both readonly sites and added them back, and this is what
I get:

# vos exa root_dist
root_dist 536871131 RW 603230 K  On-line
valhalla.icequake.net /vicepa 
RWrite  536871131 ROnly  536871132 Backup  0 
MaxQuota  0 K 
CreationSat Oct 25 22:56:52 2003
CopySun Apr 19 23:12:16 2009
Backup  Tue Jan  6 06:43:01 2004
Last Access Fri Mar 18 16:04:38 2011
Last Update Fri Mar 18 16:00:04 2011
1903 accesses in the past day (i.e., vnode references)

RWrite: 536871131 
number of sites -> 3
   server valhalla.icequake.net partition /vicepa RW Site 
   server xanadu.icequake.net partition /vicepa RO Site  -- Not released
   server valhalla.icequake.net partition /vicepa RO Site  -- Not released

# vos release root_dist
Failed to clone the RW volume 536871131
Volume already exists
Error in vos release command.
Volume already exists


-- 
Ryan C. Underwood, 
___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info

[OpenAFS] OpenAFS/Krb5/LDAP: No OpenSSH agent forwarding

2011-03-18 Thread Dirk Heinrichs 
Hi,

don't wether this fits here or not, but since I don't have this issue in
non-afs environments I guess it does.

I've got a working Debian/Ubuntu setup with OpenLDAP, MIT Kerberos5 and
OpenAFS. Upon login to one machine, an SSH agent is startet and my key
is added (via keychain). I also get kerberos tickets and an AFS token so
I can access my $HOME just fine.

However, when I now open an SSH connection to another host, I can login
without w/o any password and have immediate access to my $HOME there as
well, but the SSH agent connection is not being forwarded, although all
relevant options in sshd_config and ssh_config on the local and the
remote host are set.

That means that keychain jumps in and asks me for my SSH key passphrase
again on the remote machine. A second login to the remote machine then
works w/o any further passphrase request.

Any hints as to what could be wrong?

Thanks...

Dirk
___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info

Re: [OpenAFS] Opening files read-only when another process has a write-lock

2011-03-18 Thread Jeffrey Altman 
On 3/18/2011 12:20 PM, Thomas Smith wrote:
> On Fri, Mar 18, 2011 at 7:55 AM, Jeffrey Altman
>  wrote:
>> We are all looking forward to the day that the byte-range lock
>> implementation you have been working on can be merged into the tree.
> 
> Is there an estimate as to when this may occur?

I am hesitant to make any estimates given how long it is taking the
community to get 1.6 out the door.  I can tell you I think it will be
six months after the release of 1.10/2.0 which I think should be six to
eight months after the release of 1.6 but it is really hard to predict.




signature.asc
Description: OpenPGP digital signature

Re: [OpenAFS] Opening files read-only when another process has a write-lock

2011-03-18 Thread Thomas Smith 
On Fri, Mar 18, 2011 at 7:55 AM, Jeffrey Altman
 wrote:
> We are all looking forward to the day that the byte-range lock
> implementation you have been working on can be merged into the tree.

Is there an estimate as to when this may occur?
___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info

Re: [OpenAFS] Opening files read-only when another process has a write-lock

2011-03-18 Thread Jeffrey Altman 
On 3/18/2011 9:48 AM, Matt W. Benjamin wrote:
> Hi Thomas,
> 
> Mandatory enforcement is the defined behavior for byte-range locks on 
> Windows.  Since you are testing on Windows, you are seeing the Windows 
> behavior.  The same lock would be interpreted as advisory on a Unix client.
> 
> Matt

While this is true, this is not the problem that is preventing Office
applications from opening in-use files in read-only mode.  The problem
is that a shared lock for a particular byte-range cannot be obtained in
AFS due to the fact that the file servers do not yet support byte range
lock allocations.

We are all looking forward to the day that the byte-range lock
implementation you have been working on can be merged into the tree.

Jeffrey Altman



signature.asc
Description: OpenPGP digital signature

Re: [OpenAFS] Opening files read-only when another process has a write-lock

2011-03-18 Thread Jeffrey Altman 
On 3/18/2011 9:29 AM, Thomas Smith wrote:
> Hi,
> 
> Is it possible to open files read-only when another process has a
> write-lock? I found the following information regarding 1.5.20.
> 
> http://dl.openafs.org/dl/openafs/1.5.20/winxp/relnotes.htm

The release notes for the version of the OpenAFS client that is
installed on the machine in question is available locally

  Start->Programs->OpenAFS->Documentation->Release Notes

> Many applications on Windows (e.g. Microsoft Office) require
> the use of byte range locks
> applied to a file either to protect against simultaneous file
> access or as a signaling mechanism.
> OpenAFS for Windows release 1.5 (or greater) implements byte
> range locking within the
> CIFS-AFS gateway server.   This support for byte range locking
> obtains AFS’ advisory file
> server locks to simulate Microsoft Windows mandatory locks.
> When an application opens
> a file, a lock will be obtained from AFS indicating that the
> file is in use.  If the lock is a write
> lock, access to the file will be restricted to other
> applications running on the same machine
> as the first application to request the lock. Applications
> running on other machines will see
> the AFS full file lock and will be unable to access the file.
> 
> Just wasn't sure if this is the current state things.

It is.

> I'm on 1.4.12 on the server and 1.5.78 on the client.
> 
> This issue arose when one of my users was editing a spreadsheet and
> another user attempted to open the same spreadsheet for viewing--the
> attempted viewer received an error indicating that the file was
> locked.

As described above, Microsoft Office applications use byte range locks
as a form of signaling.  When an Office application requests a file be
opened in read-only mode it obtains a shared byte-range lock on a
particular section of the file.  That permits other instances of Office
on the network to know that the file is open for read-only use and it
should not be changed.

This model works just fine on AFS if both instances are sitting behind
the same AFS cache manager since the cache manager can manage the byte
range lock allocations.  However, it fails if the instances are on
different machines because the AFS file servers do not support
byte-range locking.

Jeffrey Altman




signature.asc
Description: OpenPGP digital signature

Re: [OpenAFS] forcing and benchmarking the cache scan

2011-03-18 Thread Derrick Brashear 
/etc/init.d/openafs-client stop; rm /usr/vice/etc/CacheItems;
/etc/init.d/openafs-client start
?


On Fri, Mar 18, 2011 at 10:05 AM, Edgecombe, Jason wrote:

> Hi everyone,
>
>
>
> I had a server doing a cache scan for over an hour yesterday after
> upgrading the RAID (mpt2sas) driver on RHEL5.5 64bit. I thought the box
> hung, so I ran a hard reset and ended up rebuilding it. I’m trying to
> reproduce and measure the cache scan time with the two different versions of
> the RAID driver to see if the driver causes an issue, or not. How can I
> force the client to run a cache scan when I run /etc/init.d/openafs-client
> restart instead of at boot time?
>
>
>
> I’m trying the following procedure to benchmark things:
>
>
>
> fs setcachesize 1
>
> fs setcachesize 62048980
>
> fs getcachep
>
> cd /afs/(some folder)
>
> find . -type f | xargs cat > /dev/null
>
> cd /
>
> time /etc/init.d/openafs-client restart
>
>
>
> Details on production box:
>
> OS: RedHat Enterprise Linux Server 5.5 (64bit)
>
> Hardware: Dell PowerEdge R210 With H200 RAID controller
>
> Driver: mp2tsas driver from Dell
>
>
>
> Details on testing box (my only H200 cards are in production servers):
>
> OS: RedHat Enterprise Linux Server 5.5 (64bit)
>
> Hardware: Dell PowerEdge R510 With H700 RAID controller
>
> Driver: mp2tsas driver from Dell
>
>
>
> Thanks,
>
> Jason
>
>
>
> ---
>
> Jason Edgecombe *| *Linux and Solaris Administrator
>
> UNC Charlotte *| *The William States Lee College of Engineering
>
> 9201 University City Blvd. *| *Charlotte, NC 28223-0001
>
> jwedg...@uncc.edu *| *http://coe.uncc.edu | [image: 
> facebook-logo]
>  Facebook 
>
> ---
>
> If you are not the intended recipient of this transmission or a person
> responsible for delivering it to the intended recipient, any disclosure,
> copying, distribution, or other use of any of the information in this
> transmission is strictly prohibited. If you have received this transmission
> in error, please notify me immediately by reply e-mail or by telephone at
> 704-687-3514.  Thank you.
>
>
>



-- 
Derrick
<>

[OpenAFS] forcing and benchmarking the cache scan

2011-03-18 Thread Edgecombe, Jason 
Hi everyone,

 

I had a server doing a cache scan for over an hour yesterday after
upgrading the RAID (mpt2sas) driver on RHEL5.5 64bit. I thought the box
hung, so I ran a hard reset and ended up rebuilding it. I'm trying to
reproduce and measure the cache scan time with the two different
versions of the RAID driver to see if the driver causes an issue, or
not. How can I force the client to run a cache scan when I run
/etc/init.d/openafs-client restart instead of at boot time?

 

I'm trying the following procedure to benchmark things:

 

fs setcachesize 1

fs setcachesize 62048980

fs getcachep

cd /afs/(some folder)

find . -type f | xargs cat > /dev/null

cd /

time /etc/init.d/openafs-client restart

 

Details on production box:

OS: RedHat Enterprise Linux Server 5.5 (64bit)

Hardware: Dell PowerEdge R210 With H200 RAID controller

Driver: mp2tsas driver from Dell

 

Details on testing box (my only H200 cards are in production servers):

OS: RedHat Enterprise Linux Server 5.5 (64bit)

Hardware: Dell PowerEdge R510 With H700 RAID controller

Driver: mp2tsas driver from Dell

 

Thanks,

Jason

 


---

Jason Edgecombe | Linux and Solaris Administrator

UNC Charlotte | The William States Lee College of Engineering

9201 University City Blvd. | Charlotte, NC 28223-0001

jwedg...@uncc.edu   | http://coe.uncc.edu
  |     
Facebook  


---

If you are not the intended recipient of this transmission or a person
responsible for delivering it to the intended recipient, any disclosure,
copying, distribution, or other use of any of the information in this
transmission is strictly prohibited. If you have received this
transmission in error, please notify me immediately by reply e-mail or
by telephone at 704-687-3514.  Thank you.

 

<>

Re: [OpenAFS] Opening files read-only when another process has a write-lock

2011-03-18 Thread Matt W. Benjamin 
Hi Thomas,

Mandatory enforcement is the defined behavior for byte-range locks on Windows.  
Since you are testing on Windows, you are seeing the Windows behavior.  The 
same lock would be interpreted as advisory on a Unix client.

Matt

- "Thomas Smith"  wrote:

> Hi,
> 
> Is it possible to open files read-only when another process has a
> write-lock? I found the following information regarding 1.5.20.
> 
> http://dl.openafs.org/dl/openafs/1.5.20/winxp/relnotes.htm
> 
> Many applications on Windows (e.g. Microsoft Office) require
> the use of byte range locks
> applied to a file either to protect against simultaneous file
> access or as a signaling mechanism.

> 
> ~ Tom
> ___
> OpenAFS-info mailing list
> OpenAFS-info@openafs.org
> https://lists.openafs.org/mailman/listinfo/openafs-info

-- 

Matt Benjamin

The Linux Box
206 South Fifth Ave. Suite 150
Ann Arbor, MI  48104

http://linuxbox.com

tel. 734-761-4689
fax. 734-769-8938
cel. 734-216-5309
___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info

[OpenAFS] Opening files read-only when another process has a write-lock

2011-03-18 Thread Thomas Smith 
Hi,

Is it possible to open files read-only when another process has a
write-lock? I found the following information regarding 1.5.20.

http://dl.openafs.org/dl/openafs/1.5.20/winxp/relnotes.htm

Many applications on Windows (e.g. Microsoft Office) require
the use of byte range locks
applied to a file either to protect against simultaneous file
access or as a signaling mechanism.
OpenAFS for Windows release 1.5 (or greater) implements byte
range locking within the
CIFS-AFS gateway server.   This support for byte range locking
obtains AFS’ advisory file
server locks to simulate Microsoft Windows mandatory locks.
When an application opens
a file, a lock will be obtained from AFS indicating that the
file is in use.  If the lock is a write
lock, access to the file will be restricted to other
applications running on the same machine
as the first application to request the lock. Applications
running on other machines will see
the AFS full file lock and will be unable to access the file.

Just wasn't sure if this is the current state things.

I'm on 1.4.12 on the server and 1.5.78 on the client.

This issue arose when one of my users was editing a spreadsheet and
another user attempted to open the same spreadsheet for viewing--the
attempted viewer received an error indicating that the file was
locked.

~ Tom
___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info

Re: [OpenAFS] Re: lock screen renew afs ticket issues?

2011-03-18 Thread Simon Wilkinson 


On 18 Mar 2011, at 10:27, TIARA System Man wrote:


sorry to bother you. now, it is fixed by restoring the previous
kerberos backup. however, i still don't know what is the correct
procedure to add another afs file server with afs key?


You have to have the same afs key on every file and database server  
within your cell. The simplest way to achieve this is to copy the  
KeyFile over from an existing server onto your new machine.


Cheers,

Simon.


___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info

[OpenAFS] Re: lock screen renew afs ticket issues?

2011-03-18 Thread TIARA System Man 
sorry to bother you. now, it is fixed by restoring the previous
kerberos backup. however, i still don't know what is the correct
procedure to add another afs file server with afs key?

thank you all.
best, sam

On Fri, Mar 18, 2011 at 4:41 PM, TIARA System Man
 wrote:
> dear all,
>
> recently, i added another afs file server into our environment.
> somehow, my users have problems after i added the new afs file server.
> as long as they they log in from lock screen, they will have the
> *permission denied* problem. what i found is before the screen saver
> triggered. they have afs/tiara.sinica.edu...@tiara.sinica.edu.tw
> ticket. after login from lock screen. the previous afs/tiara ticket is
> gone. they only has a...@tiara.sinica.edu.tw ticket.
>
> so, users need to type aklog to retrive the
> afs/tiara.sinica.edu...@tiara.sinica.edu.tw ticket. otherwise, they
> can't use the desktop.
>
> do you have any suggestion of this issue?  i'm looking for your input.
> thank you in advance.
>
> btw, my users' home directory is openafs..
>
> best, sam
>
>
> --
> Sam Tseng, Contract Assistant
> Academia Sinica Institute of Astronomy and Astrophysics
> PO BOX 23-141, TAIPEI, 10617, TAIWAN
> Tel: +886-2-2366-5392
> Fax: +886-2-2367-7849
>



-- 
Sam Tseng, Contract Assistant
Academia Sinica Institute of Astronomy and Astrophysics
PO BOX 23-141, TAIPEI, 10617, TAIWAN
Tel: +886-2-2366-5392
Fax: +886-2-2367-7849
___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info

[OpenAFS] lock screen renew afs ticket issues?

2011-03-18 Thread TIARA System Man 
dear all,

recently, i added another afs file server into our environment.
somehow, my users have problems after i added the new afs file server.
as long as they they log in from lock screen, they will have the
*permission denied* problem. what i found is before the screen saver
triggered. they have afs/tiara.sinica.edu...@tiara.sinica.edu.tw
ticket. after login from lock screen. the previous afs/tiara ticket is
gone. they only has a...@tiara.sinica.edu.tw ticket.

so, users need to type aklog to retrive the
afs/tiara.sinica.edu...@tiara.sinica.edu.tw ticket. otherwise, they
can't use the desktop.

do you have any suggestion of this issue?  i'm looking for your input.
thank you in advance.

btw, my users' home directory is openafs..

best, sam


-- 
Sam Tseng, Contract Assistant
Academia Sinica Institute of Astronomy and Astrophysics
PO BOX 23-141, TAIPEI, 10617, TAIWAN
Tel: +886-2-2366-5392
Fax: +886-2-2367-7849
___
OpenAFS-info mailing list
OpenAFS-info@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-info