toasters April 2005

toasters@lists.teaparty.net

59 participants
67 discussions

RE: File Sharing Windows <> NFS Volume on R150 (6.5.2R1) ...
by Sphar, Mike 18 Apr '05

18 Apr '05

A lot of file-locking is dependent on the client, of course. Assuming that an NFS file-lock was put in place, it's still up to the client to decide "Okay, that file is locked for writing; I will just open it up in read-only mode instead." That's my understanding anyway...I've never found nfs-based file locking to be terribly reliable. -- Michael W. Sphar - IS&T - Lead Systems Administrator SMBU Engineering Support Services, BMC Software -----Original Message----- From: McCarthy, Tim [mailto:timothy.mccarthy@netapp.com] Sent: Monday, April 18, 2005 4:27 PM To: Doug Chomyn; toasters(a)mathworks.com Subject: RE: File Sharing Windows <> NFS Volume on R150 (6.5.2R1) ... How about writing a wrapper around the file calls that does very proper file locking? I know....more work...just an idea. --tmac -----Original Message----- From: Doug Chomyn [mailto:chomyn@corefa.com] Sent: Monday, April 18, 2005 3:55 PM To: toasters(a)mathworks.com Subject: Re: File Sharing Windows <> NFS Volume on R150 (6.5.2R1) ... Um, I'm looking at dynamic sharing, where if host 1 has rw, hosts 2, 3, 4 have r only. when host 1 relinquishes w, the hosts 2, 3, 4 have the possibility of opening the file for w since they too have the privilege. McCarthy, Tim wrote: >Why not: >(6.5.1 or higher) >exportfs -p ro=host1:host2:host3,rw=host4:netgroup1:subnet/24 >/vol/vol1/path > >--tmac > >-----Original Message----- >From: Doug Chomyn [mailto:chomyn@corefa.com] >Sent: Monday, April 18, 2005 2:24 PM >To: toasters(a)mathworks.com >Subject: Re: File Sharing Windows <> NFS Volume on R150 (6.5.2R1) ... > >Yes, we have SFU installed on the Windows clients and can easily access >the NFS volumes, but I need the "file share" as if the shared document >was on a Windows NT volume, permissions-wise that is ... > >emilio brambilla wrote: > > > >>hello, >> >>Doug Chomyn wrote: >> >> >> >>>document R only, etc, etc. Ordinarily we would do a re-share via >>>Samba, or perhaps make a CIFS volume share, but I'm trying to reduce >>>complexity (and we're not going to buy a CIFS license) ... anyone >>>have experience >>> >>> >>if you have only a few workstations accesing the storage with windows >>sharing you can try installing on each workstation the "microsoft nfs >>client" you can download from http://www.microsoft.com/windows/sfu/ >> >>we did use this workaround 6 month ago to allow a few windows box to >>access a nfs-only fas270C (the customer has 300 unix boxes and only 2 >>windows clients so he did not agree to buy the cifs license) >> >>it's still working and the customer is habby about it, but we had to >>force the workstation to use nfs over tcp in order to have a stable >>connection >> >>if you need it, this is the registry key you have to set to force tcp >>connection: >>[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Client for >>NFS\CurrentVersion\Default] "PreferTCP"=dword:00000001 >> >> >> > > > >

1 0

RE: File Sharing Windows <> NFS Volume on R150 (6.5.2R1) ...
by McCarthy, Tim 18 Apr '05

18 Apr '05

How about writing a wrapper around the file calls that does very proper file locking? I know....more work...just an idea. --tmac -----Original Message----- From: Doug Chomyn [mailto:chomyn@corefa.com] Sent: Monday, April 18, 2005 3:55 PM To: toasters(a)mathworks.com Subject: Re: File Sharing Windows <> NFS Volume on R150 (6.5.2R1) ... Um, I'm looking at dynamic sharing, where if host 1 has rw, hosts 2, 3, 4 have r only. when host 1 relinquishes w, the hosts 2, 3, 4 have the possibility of opening the file for w since they too have the privilege. McCarthy, Tim wrote: >Why not: >(6.5.1 or higher) >exportfs -p ro=host1:host2:host3,rw=host4:netgroup1:subnet/24 >/vol/vol1/path > >--tmac > >-----Original Message----- >From: Doug Chomyn [mailto:chomyn@corefa.com] >Sent: Monday, April 18, 2005 2:24 PM >To: toasters(a)mathworks.com >Subject: Re: File Sharing Windows <> NFS Volume on R150 (6.5.2R1) ... > >Yes, we have SFU installed on the Windows clients and can easily access >the NFS volumes, but I need the "file share" as if the shared document >was on a Windows NT volume, permissions-wise that is ... > >emilio brambilla wrote: > > > >>hello, >> >>Doug Chomyn wrote: >> >> >> >>>document R only, etc, etc. Ordinarily we would do a re-share via >>>Samba, or perhaps make a CIFS volume share, but I'm trying to reduce >>>complexity (and we're not going to buy a CIFS license) ... anyone >>>have experience >>> >>> >>if you have only a few workstations accesing the storage with windows >>sharing you can try installing on each workstation the "microsoft nfs >>client" you can download from http://www.microsoft.com/windows/sfu/ >> >>we did use this workaround 6 month ago to allow a few windows box to >>access a nfs-only fas270C (the customer has 300 unix boxes and only 2 >>windows clients so he did not agree to buy the cifs license) >> >>it's still working and the customer is habby about it, but we had to >>force the workstation to use nfs over tcp in order to have a stable >>connection >> >>if you need it, this is the registry key you have to set to force tcp >>connection: >>[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Client for >>NFS\CurrentVersion\Default] "PreferTCP"=dword:00000001 >> >> >> > > > >

1 0

RE: File Sharing Windows <> NFS Volume on R150 (6.5.2R1) ...
by McCarthy, Tim 18 Apr '05

18 Apr '05

Why not: (6.5.1 or higher) exportfs -p ro=host1:host2:host3,rw=host4:netgroup1:subnet/24 /vol/vol1/path --tmac -----Original Message----- From: Doug Chomyn [mailto:chomyn@corefa.com] Sent: Monday, April 18, 2005 2:24 PM To: toasters(a)mathworks.com Subject: Re: File Sharing Windows <> NFS Volume on R150 (6.5.2R1) ... Yes, we have SFU installed on the Windows clients and can easily access the NFS volumes, but I need the "file share" as if the shared document was on a Windows NT volume, permissions-wise that is ... emilio brambilla wrote: > hello, > > Doug Chomyn wrote: > >> document R only, etc, etc. Ordinarily we would do a re-share via >> Samba, or perhaps make a CIFS volume share, but I'm trying to reduce >> complexity (and we're not going to buy a CIFS license) ... anyone >> have experience > > > if you have only a few workstations accesing the storage with windows > sharing you can try installing on each workstation the "microsoft nfs > client" you can download from http://www.microsoft.com/windows/sfu/ > > we did use this workaround 6 month ago to allow a few windows box to > access a nfs-only fas270C (the customer has 300 unix boxes and only 2 > windows clients so he did not agree to buy the cifs license) > > it's still working and the customer is habby about it, but we had to > force the workstation to use nfs over tcp in order to have a stable > connection > > if you need it, this is the registry key you have to set to force tcp > connection: > [HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Client for > NFS\CurrentVersion\Default] "PreferTCP"=dword:00000001 >

3 2

File Sharing Windows <> NFS Volume on R150 (6.5.2R1) ...
by Doug Chomyn 18 Apr '05

18 Apr '05

Sorry if this a bit lame. Basically, I want to file share .xls documents stored on an NFS volume between Windows clients so that when one client with RW on the document has access, another Windows client can open the document R only, etc, etc. Ordinarily we would do a re-share via Samba, or perhaps make a CIFS volume share, but I'm trying to reduce complexity (and we're not going to buy a CIFS license) ... anyone have experience with doing this kind of thing? Is it possible? tx doug

2 2

Re: Converting to DP
by Reinoud Reynders 15 Apr '05

15 Apr '05

Hi Robert, It's just the same as a normal reconstruction when a disk fails. The time depends on the disk size, filer and the value of your reconstruction speed (low, middle, high). Best regards, Reinoud Reynders UZLeuven Belgium -----Original Message----- From: owner-toasters(a)mathworks.com <owner-toasters(a)mathworks.com> To: Netapp <toasters(a)mathworks.com> Sent: Fri Apr 15 17:55:48 2005 Subject: Converting to DP I have a 13 disk volume that is 600 Gig in size. I also made the raid group 13 disks originally when I first got the Filer in. Its a FAS 940 running 6.5.4. With 56 (72) gig disks 13 disk volumes divided out nicely after subtracting 2 for vol0 and 2 for spares. Now that I'm running 6.5 its time for DP. I've got 3 spares now and want to convert the 600 Gig volume to DP. I upped the raid group on the volume to 14 and plan to issue the command to DP Sunday morning. The volume is 60% full. We run Oracle on it but all DB's will be down Sunday at 7AM. How long will the convet take? I understand the entire volume's parity will need to be recomputed yes? Anybody have a similar environment and situation? TIA -Rob

1 0

RE: Changing Netapp SCSI target ID
by McCarthy, Tim 15 Apr '05

15 Apr '05

You cannot have more than one filer attached to the same SCSI chain. What you can do is have one drive attached to one filer and another drive attached to another filer, as long as you can break the SCSI bus at the library side. Most libraries with more than 2 drives can do this. You will see multiple in/out connections. That is what it is for. --tmac ====================== Tim McCarthy Professional Services/Systems Engineer NetApp Federal Systems, Inc. 410-551-3970 (o) 443-363-0208 (f) tmac(a)netapp.com tmac-pager(a)netapp.com ====================== -----Original Message----- From: Alan Sparks [mailto:asparks@doublesparks.net] Sent: Friday, April 15, 2005 11:00 AM To: Lance R Bailey Cc: Toasters List Subject: Re: Changing Netapp SCSI target ID Lance R Bailey said: > i have a qualstar library with the drives attached to a linux host, > F720, F87 and F85 filers. > > the scsi number is set on the library. each drive slot has a different > number. when the filer sees it (verify with sysconfig -t) the scsi > number assigned to the drive on the scsi bus will be the one on > library. OK. So you're saying that it will not cause a conflict if I daisy chain the filers together onto the same wire to the library drives. OK, that or 3-way NDMP are two possible options then. Thanks. -Alan =========== Alan Sparks, UNIX/Linux Systems Administrator <asparks(a)doublesparks.net>

3 2

Re: ndmp
by Jerry 15 Apr '05

15 Apr '05

I'm not sure I responded to it, but here is the bug: http://now.netapp.com/NOW/cgi-bin/bol?Type=Detail&Display=96894 Bug 96894 Fixed in 6.4.2P4 and higher. --- George Kahler <george(a)YorkU.CA> wrote: > Few weeks ago someone has replied to a post about > ndmp being very > slow in its phase one. The person replied that this > has been fixed > now in a particular version of Data ONTAP. > > I deleted the message and now I would like to find > out what version of > ONTAP the person said it was fixed in. > > Could whoever sent it please resent it to me ? > > Thanks, > George > __________________________________ Do you Yahoo!? Yahoo! Small Business - Try our new resources site! http://smallbusiness.yahoo.com/resources/

1 0

Converting to DP
by Robert Borowicz 15 Apr '05

15 Apr '05

I have a 13 disk volume that is 600 Gig in size. I also made the raid group 13 disks originally when I first got the Filer in. Its a FAS 940 running 6.5.4. With 56 (72) gig disks 13 disk volumes divided out nicely after subtracting 2 for vol0 and 2 for spares. Now that I'm running 6.5 its time for DP. I've got 3 spares now and want to convert the 600 Gig volume to DP. I upped the raid group on the volume to 14 and plan to issue the command to DP Sunday morning. The volume is 60% full. We run Oracle on it but all DB's will be down Sunday at 7AM. How long will the convet take? I understand the entire volume's parity will need to be recomputed yes? Anybody have a similar environment and situation? TIA -Rob

1 0

Re: RAID-DP
by George Kahler 15 Apr '05

15 Apr '05

Thanks Adam and to others for clearing up the misconception. George On Mon, 4 Apr 2005 09:17:09 -0700, "Fox, Adam" <Adam.Fox(a)netapp.com> wrote: > This is absolutely false. I heard this strange misconception with > RAID-4 too. > (If you lose the parity disk, the whole thing falls down because it's > all on 1 disk). > > Here's a lab box to make you feel better. > > Volume test (online, raid_dp) (block checksums) > Plex /test/plex0 (online, normal, active) > RAID group /test/plex0/rg0 (normal) > > RAID Disk Device HA SHELF BAY CHAN Used (MB/blks) Phys > (MB/blks) > --------- ------ --------------- ---- -------------- > -------------- > dparity 0b.20 0b 1 4 FC:B 68000/139264000 > 69536/142410400 > parity 0b.21 0b 1 5 FC:B 68000/139264000 > 69536/142410400 > data 0b.28 0b 1 12 FC:B 68000/139264000 > 69536/142410400 > > > Spare disks (empty) > > (note: No spares) > > filer> disk fail -i 0b.20 > filer> disk fail -i 0b.21 > > Now, here's sysconfig -r again: > > Volume test (online, raid_dp, degraded) (block checksums) > Plex /test/plex0 (online, normal, active) > RAID group /test/plex0/rg0 (double degraded) > > RAID Disk Device HA SHELF BAY CHAN Used (MB/blks) Phys > (MB/blks) > --------- ------ --------------- ---- -------------- > -------------- > dparity FAILED N/A 68000/139264000 > parity FAILED N/A 68000/139264000 > data 0b.28 0b 1 12 FC:B 68000/139264000 > 69536/142410400 > > > Spare disks (empty) > > Broken disks > > RAID Disk Device HA SHELF BAY CHAN Used (MB/blks) Phys > (MB/blks) > --------- ------ --------------- ---- -------------- > -------------- > admin failed 0b.20 0b 1 4 FC:B 68000/139264000 > 69536/142410400 > admin failed 0b.21 0b 1 5 FC:B 68000/139264000 > 69536/142410400 > > Still running..but degraded. > > -- Adam Fox > adamfox(a)netapp.com > > > > -----Original Message----- > From: George Kahler [mailto:george@YorkU.CA] > Sent: Monday, April 04, 2005 11:22 AM > To: toasters(a)mathworks.com > Subject: RAID-DP > > At a recent NetApp presentation it was mention in passing that RAID-DP > fails if a double disk failure happens to hit the two parity disk. > > Is this true ? > > George

1 1

Changing Netapp SCSI target ID
by Alan Sparks 15 Apr '05

15 Apr '05

Possibly a weird question, but... I need to connect a Qualstore library to my toasters (760 filers) by daisy-chaining them to the library. Just don't have the ports on the library to direct connect to the drives. But unless I really misunderstand, these filers will all have the same SCSI IDs, so there's gonna be a conflict. I need to let my NetBackup server control everything, and run NDMP on the filers to do the backups. Is there any way to retarget the SCSI ports built into the Netapp head to be on an ID of my choosing, so I can sidestep the ID conflict? Thanks in advance. -Alan -- Alan Sparks, UNIX/Linux Systems Integration and Administration <asparks(a)doublesparks.net>

2 3

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 April 2005