toasters

toasters@lists.teaparty.net

3 participants
13514 discussions

RE: Interesting permissions problem
by Muhlestein, Mark 12 Jul '99

12 Jul '99

OK, try going through the security troubleshooter: http://now.netapp.com/knowledge/docs/olio/guides/53_troubleshooting/index.s… If the qtree allows ACLs (i.e. it is not a UNIX qtree) there may be an ACL on the file that does not grant you access. In that case you are actually looking at the display permissions from the UNIX side. Let me know what you find. Mark -----Original Message----- From: Graham C. Knight [mailto:grahamk@ast.lmco.com] Sent: Monday, July 12, 1999 9:08 AM To: Muhlestein, Mark Cc: 'toasters(a)mathworks.com' Subject: Re: Interesting permissions problem The file does not appear in a lock_dump listing. Here is some interesting behaviour: [diamond] ~/ups 3> ls -al total 1000 drwx------ 2 kfackler nt 4096 Jul 8 09:11 ./ drwx------ 4 kfackler nt 4096 Jul 12 08:41 ../ -rwx------ 1 kfackler nt 22528 Jun 24 08:34 UPS Review.doc* -rwx------ 1 kfackler nt 19968 Jul 8 09:10 UPS Server Backup.doc* -rwx------ 1 kfackler nt 202752 Jul 6 09:09 UPStempl.mpp* -rwx------ 1 kfackler nt 194560 Jul 6 09:12 UPStempl.mpt* -rwx------ 1 kfackler nt 45568 Jun 29 09:41 WhoOpsWhat.doc* [diamond] ~/ups 4> whoami kfackler [diamond] ~/ups 5> chmod 700 * chmod: WARNING: can't change UPS Server Backup.doc [diamond] ~/ups 6> more "UPS Server Backup.doc" UPS Server Backup.doc: Permission denied [diamond] ~/ups 7> I cannot do anything with this file - it doesn't appear to matter who i am. I cannot do anything with this file from NT when i'm logged in as a user who has a UID of zero either. It won't let me copy it, or see the permissions. It gives me the "make sure the disk isn't full" error. Any tips appreciated.... Thx, Graham "Muhlestein, Mark" wrote: > > It looks suspiciously like a CIFS lock. There is a hidden command "lock_dump -f" > which is available after doing "rc_toggle_basic". It shows all the cifs locks. > > The reason the files are locked even for root is that SMB locks are mandatory. > We are following NT semantics. It is not perfectly clear that you would always > want to strictly follow NT semantics in all cases, and we are looking at > addressing that issue. > > If you can do your backups from snapshots you will avoid the problem of files > locked by CIFS clients. > > Mark Muhlestein -- mmm(a)netapp.com > > -----Original Message----- > From: Graham C. Knight [mailto:grahamk@ast.lmco.com] > Sent: Monday, July 12, 1999 7:58 AM > To: toasters(a)mathworks.com > Subject: Interesting permissions problem > > A filer of ours (ast-h-dcf04) has NFS exported as > root to a system named "nimbus". The filer is > giving permission denied messages when trying to > access some files as root. This is causing backups > to fail for this file (and a few others). Anyone > seen this type of behaviour?: > > [nimbus] ~ 103> sudo file > "/data/ast-h-dcf04/ast-h-dcf04-1/kfackler/ups/UPS Server Backup.doc" > /data/ast-h-dcf04/ast-h-dcf04-1/kfackler/ups/UPS Server Backup.doc: > cannot open: Permission denied > > Other files in the same directory can be accessed, and they > have identical permissions as this one: > > -rwx------ 1 kfackler nt 19968 Jul 8 09:10 UPS Server > Backup.doc* > > I thought maybe this user had it locked in CIFS, but: > > ast-h-dcf04> cifs sessions kfackler > > users > shares/files opened > > MA106734 (ACCT02\kfackler - kfackler) > {home_dir} > > Is there a way to list the files > locked by CIFS on a filer? And even > if it were locked - why can't root > do querires on the file, or back it up? > > This filer is running 5.3D14. Hopefully > I'm missing something very obvious here.... > > Thx, > Graham

1 0

New shelves on 330
by Jay Orr 12 Jul '99

12 Jul '99

Hi, Ok, so we had gotten an old shelf from our other branch and we tried to bring it online sunday. It seems that the old shelf was the boot shelf (off of controller 0) and we put it on scsi 9b on our system. When the system was rebooted it crashed into PROM mode with the error message raid ID error 0,5. We did this twice and it returned the same error with different drive descriptors both times (but both 0,5 ID error). Anyone know what to do with this? ----------- Jay Orr

1 1

RE: Interesting permissions problem
by Muhlestein, Mark 12 Jul '99

12 Jul '99

It looks suspiciously like a CIFS lock. There is a hidden command "lock_dump -f" which is available after doing "rc_toggle_basic". It shows all the cifs locks. The reason the files are locked even for root is that SMB locks are mandatory. We are following NT semantics. It is not perfectly clear that you would always want to strictly follow NT semantics in all cases, and we are looking at addressing that issue. If you can do your backups from snapshots you will avoid the problem of files locked by CIFS clients. Mark Muhlestein -- mmm(a)netapp.com -----Original Message----- From: Graham C. Knight [mailto:grahamk@ast.lmco.com] Sent: Monday, July 12, 1999 7:58 AM To: toasters(a)mathworks.com Subject: Interesting permissions problem A filer of ours (ast-h-dcf04) has NFS exported as root to a system named "nimbus". The filer is giving permission denied messages when trying to access some files as root. This is causing backups to fail for this file (and a few others). Anyone seen this type of behaviour?: [nimbus] ~ 103> sudo file "/data/ast-h-dcf04/ast-h-dcf04-1/kfackler/ups/UPS Server Backup.doc" /data/ast-h-dcf04/ast-h-dcf04-1/kfackler/ups/UPS Server Backup.doc: cannot open: Permission denied Other files in the same directory can be accessed, and they have identical permissions as this one: -rwx------ 1 kfackler nt 19968 Jul 8 09:10 UPS Server Backup.doc* I thought maybe this user had it locked in CIFS, but: ast-h-dcf04> cifs sessions kfackler users shares/files opened MA106734 (ACCT02\kfackler - kfackler) {home_dir} Is there a way to list the files locked by CIFS on a filer? And even if it were locked - why can't root do querires on the file, or back it up? This filer is running 5.3D14. Hopefully I'm missing something very obvious here.... Thx, Graham

2 1

Re: New shelves on 330
by Dave Atkin 12 Jul '99

12 Jul '99

> Ok, so we had gotten an old shelf from our other branch and we tried to > bring it online sunday. It seems that the old shelf was the boot shelf > (off of controller 0) and we put it on scsi 9b on our system. When the > system was rebooted it crashed into PROM mode with the error message raid > ID error 0,5. We did this twice and it returned the same error with > different drive descriptors both times (but both 0,5 ID error). Anyone > know what to do with this? If it's a storageshelf 2, there's a DIP switch on the back that needs to be set. The hardware manual says that for the second shelf the switches (numbered 1-8) should be off,off,off,on,off,off,on,on. Hope this helps. Dave Atkin ------------------------------------------------------ Dave Atkin, Head of Technical Services Computing Service, University of York, YORK YO10 5DD Phone: +44-1904-433804 (ddi) Fax: +44-1904-433740 Email: D.Atkin(a)york.ac.uk ------------------------------------------------------

1 0

ufsdump/ufsrestore and Solaris 2.7
by Dave Atkin 12 Jul '99

12 Jul '99

Sun seem to have "improved" ufsdump/ufsrestore in Solaris 2.7 and they are now so good that they won't talk to a tape drive on a filer. The two utilities now seem to require a feature of the RMT protocol that NetApp don't bother to implement - the "S" (status) command. I've reported the problem to NetApp and they are going to add it to their wish list of enhancements. In the meantime a workaround seems to be to copy over the 2.6 binaries of ufsdump/ufsrestore which seem to work better. Dave Atkin ------------------------------------------------------ Dave Atkin, Head of Technical Services Computing Service, University of York, YORK YO10 5DD Phone: +44-1904-433804 (ddi) Fax: +44-1904-433740 Email: D.Atkin(a)york.ac.uk ------------------------------------------------------

1 0

32-bit vs. 64-bit FC-AL adapters
by Brian Tao 10 Jul '99

10 Jul '99

I just noticed this on our clustered pair of F740's. The on-board (slot 0) adapter is listed as a 64-bit device, while the secondary adapter (slot 1, connecting to its partner's disk shelf) is listed as a 32-bit device. Is this right? NetApp Release 5.2.2: Sat Apr 17 04:15:20 PDT 1999 System ID: 0016791871 (nfs3); partner ID: 0016791820 (nfs4) slot 0: System Board Memory Size: 512 MB slot 0: FC adapter: isp2100 (chip rev. 3, 64-bit) Firmware rev: 1.15.2 Host Loop Id: 119 FC Node Name: 2:000:00a098:035eff Cacheline size: 8 FC Packet size: 2048 7 Disks: 119.0GB 1 shelf with Eurologic EDM SCSI Host Adapter 0b In-Band Enclosure Services slot 0: Ethernet Controller e0 MAC Address: 00:a0:98:00:50:46 (auto-100tx-fd-up) slot 1: FC adapter: isp2100 (chip rev. 3, 32-bit) Firmware rev: 1.15.2 Host Loop Id: 119 FC Node Name: 2:000:00e08b:00c089 Cacheline size: 8 FC Packet size: 2048 7 Disks: 119.0GB 1 shelf with Eurologic EDM slot 3: Ethernet Controller e3 MAC Address: 00:c0:95:fa:cd:7b (auto-100tx-up) slot 8: Interconnect adapter slot 9: NVRAM Memory Size: 32 MB -- Brian Tao (BT300, taob(a)risc.org) "Though this be madness, yet there is method in't"

1 0

(no subject)
by Mohler, Jeff 08 Jul '99

08 Jul '99

1 0

ForSale F330
by Wifehnter1＠aol.com 08 Jul '99

08 Jul '99

F 330 128megs Ram 2 SCSI controllers 4 meg Cache 4 drive rack consisting of 28 4.3 gig drives This unit is only 18 months old! Craig R. von Esch

2 1

Re: NIS Problems
by dnb＠ccs.neu.edu 07 Jul '99

07 Jul '99

"Sivo, Peter " <peter.sivo(a)netapp.com> writes: > Is this the reason that you all do this? I thought it was because > Corp IT never had reliable NIS services and you were binding to a > server that you knew was "blessed" and stable? On the whole, our NIS servers are rock solid. We bind to a specific IP address for the reasons Jay stated (namely we know not to trust NIS, and especially NIS broadcasts as far as we can throw it/them). > The questions, though, is that are people going to continue to hardcode in a > server, so even if I had 5 NIS servers/slaves per network, it wouldn't make a > difference? There are two separate issues here: 1) The current NetApp NIS client code (especially when run with a single server specified as we do) isn't particularly robust. If we could specify 5 IP addresses instead of one, and the failover went smoothly, this would help quite a bit. Plus you could specify hosts in a different broadcast domain (perhaps on a more protected subnet). 2) NIS will never be secure, but being able to tell a filer not to pay attention to any one who pretends to be your NIS server (via specifying addresses) is a good thing. Not the perfect thing, but better than the current status quo. Besides the scenario Jay proposed where the rogue server is available before the real one after a power blink, you could also probably throw enough of a denial-of-service against either the filer or the NIS server to cause the filer to attempt a rebind. If your filer rebinds to the rogue... To directly answer your question, it's the sixth NIS server on your net (that you didn't put there) that's the real bummer. Peace, dNb eagerly looking forward to the new NIS code

1 0

Re: NIS Problems
by Devi Nagaraj 06 Jul '99

06 Jul '99

Hi, We are revamping our NIS client implementation, one of the changes being an option to specify multiple preferred NIS servers. Please get in touch with NetApp for more information on the availability of these changes. Thanks, -Devi At 09:49 AM 7/6/99 -0600, Graham C. Knight wrote: >Sean Hinchey wrote: > >> >> Does anyone have any recommendations or are we stuck with less stable NIS >> client? >> >> -Sean > >After battling this same problem for months i finally >put some hooks into my makefile.nis that copies >the files locally to all our filers and turned >off NIS. You are correct, Netapp's are not good >NIS clients. > >Graham > >

1 0

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

toasters