I am currently running some POSIX (fstests) on a client. I noticed sometimes I get the following when the user id does not exist when issuing a chmod/chown command. The user id truly does not exist but not sure why it is causing the client to hang. Has anyone else ran into this problem? I am running openSuse 11.1 with Lustre version 1.8.1 Thanks, Stephen ___________________________________________________________________ This electronic mail message, including any attachments or embedded documents, is intended only for the addressee and may contain privileged and/or confidential information. Unless expressly indicated, this electronic mail message is confidential and privileged information of Nirvanix, Inc. Any use, dissemination, copying or distribution in any form, and any action taken based up this email is strictly prohibited unless expressly indicated. If you have received this message by error, please immediately delete it and any copies of it, including any attachments hereto, and notify the sender at Nirvanix, Inc. by reply electronic mail message or phone. Thank you. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.lustre.org/pipermail/lustre-discuss/attachments/20090921/b8673108/attachment-0001.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/png Size: 24059 bytes Desc: image001.png Url : http://lists.lustre.org/pipermail/lustre-discuss/attachments/20090921/b8673108/attachment-0001.png
On Sep 22, 2009, at 1:31 AM, Stephen Hooper wrote:> I am currently running some POSIX (fstests) on a client. I noticed > sometimes I get the following when the user id does not exist when > issuing a chmod/chown command. The user id truly does not exist but > not sure why it is causing the client to hang. Has anyone else ran > into this problem? > > I am running openSuse 11.1 with Lustre version 1.8.1This problem will be fixed in 1.8.1.1 (under testing at the moment). Meanwhile, you can apply attachment 25456 from bug 20020. Johann
Hello ALL!
I am Lustre beginner.
I built a 4nodes cluster of lustre, That is each node for OSS also for
Client. When I installed the lustre-client-xxx.rpm and
lustre-client-modules-xxx.rpm,I received some file conflit warnning . so I
installed them by option --force.
By the way, The nodes are the computing nodes also.
-------------system info------------------
OS:RHEL 5u2
Lustre Version 1.6.7.1
4nodes just as below:
panda1: MDS OSS Client
panda2: OSS Client
panda3: OSS Client
panda4: OSS Client
------------system info-------------------
Qustion1:Is it risk when the OSS and Client in one nodes? such as: lustre crash
, performance degradation when caculating
Qustion2:If lustre crash , what I cant do?
Thanks a lot~
2009-09-23
fengya0785
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.lustre.org/pipermail/lustre-discuss/attachments/20090923/651e3fb3/attachment-0001.html
I don?t believe that you need to install the client RPMs when you install the
server. I wouldn?t force install them. The client will mount with just the
server RPMs.
We have seen cases when clients will cause you to have to reboot a machine
because of bugs or other issues. When that happens, you take down a portion of
your cluster. While we have clients running on OSS machines, we do not have
anything other than the MDS running on the MDS machines. If the MDT becomes
damaged or corrupted, you can lose your entire cluster. For us, that risk was
something to be avoided.
--
Andrew
From: lustre-discuss-bounces at lists.lustre.org [mailto:lustre-discuss-bounces
at lists.lustre.org] On Behalf Of fengya0785
Sent: Wednesday, September 23, 2009 12:23 AM
To: lustre-discuss
Subject: [Lustre-discuss] Lustre error
Hello ALL!
I am Lustre beginner.
I built a 4nodes cluster of lustre, That is each node for OSS also for
Client. When I installed the lustre-client-xxx.rpm and
lustre-client-modules-xxx.rpm,I received some file conflit warnning . so I
installed them by option --force.
By the way, The nodes are the computing nodes also.
-------------system info------------------
OS:RHEL 5u2
Lustre Version 1.6.7.1
4nodes just as below:
panda1: MDS OSS Client
panda2: OSS Client
panda3: OSS Client
panda4: OSS Client
------------system info-------------------
Qustion1:Is it risk when the OSS and Client in one nodes? such as: lustre crash
, performance degradation when caculating
Qustion2:If lustre crash , what I cant do?
Thanks a lot~
2009-09-23
________________________________
fengya0785
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.lustre.org/pipermail/lustre-discuss/attachments/20090924/43a2a20b/attachment.html
On Wed, 2009-09-23 at 14:22 +0800, fengya0785 wrote:> Hello ALL!Hi,> I am Lustre beginner. > I built a 4nodes cluster of lustre, That is each node for OSS also > for Client.This is an unsupported configuration due to a theoretical memory deadlock. Some people use it, but we just like to warn about it.> When I installed the lustre-client-xxx.rpm and > lustre-client-modules-xxx.rpm,I received some file conflit warnning .What were the warnings?> so I installed them by option --force.You should *NEVER* force the installation of RPMs. If they don''t install together they either were not meant to or there is a bug which needs reporting.> > Qustion1:Is it risk when the OSS and Client in one nodes? such as: > lustre crash , performance degradation when caculatingYes. Potential deadlock due to memory pressure.> Qustion2:If lustre crash , what I cant do?Recover it? Not really sure what you are asking here. b. -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 197 bytes Desc: This is a digitally signed message part Url : http://lists.lustre.org/pipermail/lustre-discuss/attachments/20090924/1eec5d93/attachment.bin