toasters

toasters@lists.teaparty.net

13514 discussions

Re: ONTAP with RedHat linux
by Jered Heeschen 18 Oct '01

18 Oct '01

It's nice to hear I'm not the only one who's run into the hang with the 2.4 kernel, though it sounds like we get it more consistently. I've seen the same issue on a RedHat 6.2 box that's updated to the 2.4 kernel, which is why I'd point to something in the kernel rather than something RedHat-specfic. The best workaround seems to be using TCP to mount the NFS shares - that should allow for better performance than limiting readand write sizes to 1023. To do that, add "tcp" to the mount options, either in /etc/fstab or using the -o flag for "mount". You'd also need to turn on the nfs.tcp.enable option on the filer. Jered Steve Losen <scl(a)sasha.acc.virginia.edu>@mathworks.com on 10/18/2001 10:25:54 AM Sent by: owner-toasters(a)mathworks.com To: toasters(a)mathworks.com cc: Subject: Re: ONTAP with RedHat linux I am running RedHat 7.1 and NFS mounting a netapp F820 running 6.0.1R3. Things work pretty smoothly, but I have experienced the same problem as another poster (sorry, lost the message) where NFS writes occasionally delay and a "NFS server not responding" error is logged. I have not tried the rsize=1023,wsize=1023 work around. I have experienced the same problem with NFS servers other than Netapp, including a IBM RS6000 running AIX. Non Linux NFS clients are having no delay problems with the Netapp filers, so I'm pretty sure it is on the Linux end. For awhile I thought the NIC on my Linux box had botched the speed/duplex negotiation with the ethernet switch. Our network folks can tell me what speed the switch port has negotiated, but I don't know how to check my NIC from Linux. The network folks tell me they see no errors on my switch port when I experience NFS delays. I don't seem to have any other network problems, so I doubt this is the problem. I have been able to duplicate the problem on a second Linux box. However, on a third box, essentially identical to mine and also running RH 7.1, I could not duplicate it. A couple of things to be aware of with RedHat Linux and NFS. Prior to RedHat 7.0, only NFS v2 was supported. Furthermore, Linux only supported 16 bit uids and gids. We have a lot of files on our Netapp filers with uids greater than 65535. RedHat 7.1 supports NFS v3 and 32 bit uids and gids. 32 bit uids are critical in our environment. I administer our filers and when I was running RH 6.2, I could not get accurate file owner information with "ls" and I could not "chown" files to uids over 65535, so I had to use another unix box for filer administration. Now I can do it all from my Linux box. Steve Losen scl(a)virginia.edu phone: 434-924-0640 University of Virginia ITC Unix Support

1 0

RE: disk_fw_update
by Geoff Hardin 18 Oct '01

18 Oct '01

Thanks for the responses everyone! Jeff Sloan wrote: Yes there is definitely spin up issues with some of the old firmware on the 18G FCAL disks. Upgrading the firmware to the newest releases is the right move. Doing it individually is a very safe method up updating them, and definitely desirable, as you've had a few disk fail on you. I would like to note this is just on the older firmware revisions on the 18G drives. (hence once you finish upgrading you won't have to worry about this repeating itself). In regard to your current situation, The solution is to reboot the node that is reporting the mailbox disk error. We have a workaround that allows you to avoid rebooting, but it's only for ONTAP 6.1X and above. Did you halt the second system when upgrading the firmware? There's a specific process to upgrading the disk firmware on a cluster pair. You can review that process here: http://now.netapp.com/NOW/knowledge/docs/ontap/rel601r3/html/sag/fw_up5.htm… Cheers! :-) Jeff > -----Original Message----- > From: Geoff Hardin [mailto:geoff.hardin@dalsemi.com] > Sent: Thursday, October 18, 2001 1:40 PM > To: Sloan, Jeff > Subject: Re: disk_fw_update > > > Jeff, > The cluster is running DOT NetApp Release 5.3.4R2. We > are updating the firmware in anticipation of upgrading to > DOT 6.1.1R1. We have had problems with some of our filers > with the old 18GB disks during the upgrade; during the last > upgrade, we lost 3 disks, fortunately in 3 separate RAID > groups. Needless to say, it made for a very long, very > tense night as the disks rebuilt. We were hoping to update > the firmware of each disk separately so that we could avoid > this problem. > > Geoff Hardin > geoff.hardin(a)dalsemi.com > Politicians and diapers need changing often for similar reasons.

1 0

disk_fw_update
by Geoff Hardin 18 Oct '01

18 Oct '01

filer1> cf status filer1 has detected a mailbox disk error, takeover of filer2 disabled I have just finished upgrading the disk firmware on filer2. I had turned off clustering per the disk_fw_upgrade instructions prior to upgrading the disks. Once I had finished, I tried to turn clustering back on on filer1 and received the error message above. Any idea what's going on? filer2 does not report any errors on cf status. Thanks for the help! Geoff Hardin Dallas Semiconductor MIS UNIX geoff.hardin(a)dalsemi.com Politicians and diapers need changing often for similar reasons.

2 1

RE: quotas file
by Honeycutt, Michael 18 Oct '01

18 Oct '01

Thats how I do it also. For instance, if you want to use a scheduled task to do a quota off/on instead of resize, you could say: This would be first cmd file- rsh filername -l username:password -n quota off volumename call c:\foldername\filename.cmd This would be second cmd file- rsh filername -l username:password -n quota on volumename -----Original Message----- From: Jordan Share [mailto:iso9@phantasticant.com] Sent: Thursday, October 18, 2001 2:23 PM To: Grey, Michael; toasters(a)mathworks.com Subject: RE: quotas file Well, I can't help you with most of your questions, but I can tell you how I rsh from win2k to my filer. In my hosts.equiv file, I have a line like this: w.x.y.z Administrator Where w.x.y.z is the IP of the machine I am rsh'ing from. Then I use commands like this: rsh FILERNAME -l root:ROOTPASSWORD command_for_filer This works for me. Jordan -----Original Message----- From: owner-toasters(a)mathworks.com [mailto:owner-toasters@mathworks.com]On Behalf Of Grey, Michael Sent: Wednesday, October 17, 2001 8:12 PM To: 'toasters(a)mathworks.com' Subject: quotas file Well I have figured out how to set quotas based on global groups in Windows NT, but I'm having a little problem with the quotas file once completed. My script checks to see if a user is a member of a global group for a quota alocation (200MB, 350MB, and 500MB) via the "Global" command. This script runs automatically when a change is made to a quota global group. If the user ID is not in any global group then they would receive the default quota. The users who are members of a global group for quotas are written to a scratch file in the format of the quotas file. Once the script is complete the quotas file is deleted and then this scratch file is renamed to the quotas file. Everything works excellent but quota changes are not applied until I perform a "quota resize" command. I need to know if the filer automatically recalculates the quotas based on this file and if so, under what conditions? or, is there any way to have the filer recalculate quotas based on the quota file other then via the "quota resize" command. I'm trying to do this from a batch script under NT. I'm not sure how to RSH to the box via a script from NT if at all possible. Any help would be appreciated. Thanks.

1 0

RE: quotas file
by Berry, Grant 18 Oct '01

18 Oct '01

Hi Michael. Sounds like you have quotas set up well. RSH from an NT client is pretty easy to set up and then make a batch file for. This is what I do... 1) Just type in "rsh filername command" eg: rsh toronto1 df from a dos prompt of the machine that you wish to schedule this from. At the filer console you'll get an rsh error similar to: toronto1> Thu Oct 18 07:34:26 EDT [rshd_0]: Permission denied to rsh request fro m gberry at 99.118.225.53 2) Using that info for an example go and edit your <filerroot>/etc/host.equiv file with wordpad and add a line that contains "99.118.225.53 gberry" and make sure you hit <enter> at the end of the line, save the file. Then try "rsh toronto1 df" again to test and it should work just fine. A .bat file to do this would simply contain the command(s) you wish to execute. You can also use the "-l <username>" option on the rsh command if that's handy. Just type rsh at a dos prompt and it'll give you the syntax. I hope this helps. Regards, Grant -----Original Message----- From: Grey, Michael [mailto:Michael.Grey@ttc.ca] Sent: Wednesday, October 17, 2001 11:12 PM To: 'toasters(a)mathworks.com' Subject: quotas file Well I have figured out how to set quotas based on global groups in Windows NT, but I'm having a little problem with the quotas file once completed. My script checks to see if a user is a member of a global group for a quota alocation (200MB, 350MB, and 500MB) via the "Global" command. This script runs automatically when a change is made to a quota global group. If the user ID is not in any global group then they would receive the default quota. The users who are members of a global group for quotas are written to a scratch file in the format of the quotas file. Once the script is complete the quotas file is deleted and then this scratch file is renamed to the quotas file. Everything works excellent but quota changes are not applied until I perform a "quota resize" command. I need to know if the filer automatically recalculates the quotas based on this file and if so, under what conditions? or, is there any way to have the filer recalculate quotas based on the quota file other then via the "quota resize" command. I'm trying to do this from a batch script under NT. I'm not sure how to RSH to the box via a script from NT if at all possible. Any help would be appreciated. Thanks.

2 1

quotas file
by Grey, Michael 18 Oct '01

18 Oct '01

Well I have figured out how to set quotas based on global groups in Windows NT, but I'm having a little problem with the quotas file once completed. My script checks to see if a user is a member of a global group for a quota alocation (200MB, 350MB, and 500MB) via the "Global" command. This script runs automatically when a change is made to a quota global group. If the user ID is not in any global group then they would receive the default quota. The users who are members of a global group for quotas are written to a scratch file in the format of the quotas file. Once the script is complete the quotas file is deleted and then this scratch file is renamed to the quotas file. Everything works excellent but quota changes are not applied until I perform a "quota resize" command. I need to know if the filer automatically recalculates the quotas based on this file and if so, under what conditions? or, is there any way to have the filer recalculate quotas based on the quota file other then via the "quota resize" command. I'm trying to do this from a batch script under NT. I'm not sure how to RSH to the box via a script from NT if at all possible. Any help would be appreciated. Thanks.

2 1

Ndmpcopy or vol copy (data migration)?
by Dawn Lovell 18 Oct '01

18 Oct '01

I know this very issue has been discussed previously, but I'm seeking some clarification. We have a 210 running 5.3.6R2 that we're replacing with an 820 running 6.1.1R1. The filers both have 100mb/full-duplex connections on the same subnet. Our initial plan was to vol copy the single 36GB volume on the 210 to a new volume on the 820. We've used vol copy to move roughly the same amount of data from a 210 to a 740 in the past, although both were running the same DOT version. After reading the archived discussion on the filesystem reversion necessary when migrating between 5.3.6R2 and 6.0.1R1, I opened a case with NetApp to ask about the time involved for reversion and if the filesystem would be upgraded after the vol copy completed. The support engineer pointed out that there are possible data safety issues with doing a vol copy from the older hardware/OS to the newer and strongly suggested avoiding vol copy and using ndmpcopy instead. I guess I'm just seeking some additional opinions, as vol copy has worked so well for us in the past. We can certainly use ndmpcopy with some variation of level0/incremental (or rsync), but I'm concerned the move will end up taking longer that way than if we just schedule several hours of downtime and vol copy once. I don't want to do something unsupported, but my impression (from the archived discussion) was that vol copy should work after the target filesystem was reverted. Is vol copy actually unsupported by NetApp in this scenario? Has anyone done this and had problems? Thanks very much for your time! Dawn Lovell dlovell(a)centurytel.net

1 0

RE: root volume
by Lane, Debbie 18 Oct '01

18 Oct '01

This list is great. I wanted to thank everyone for their replies. I was also impressed with NetApp Tech Support. They saw my question on the list server and gave me a call. Best Regards, --Debbie

1 0

RE: NetApp and EMC Comparisons
by Joe Luchtenberg 18 Oct '01

18 Oct '01

1. The article, while making some valid points about the industry in general with regards to FC SAN and TCP/IP NAS, misses a lot of significant points about Network Appliance. Furthermore, in making generalized statements about the NAS market (most platforms of which are tuned Win2K Advanced Server devices or some-flavor-of-Linux), these would mis-characterize filers if applied to them. When comparing NetApp filers to competing SANs the important comparisons to make are between WAFL and NTFS/xFS. For instance, when addressing the scalability requirements for large enterprises, The Clipper Group states, "[NAS] will likely require, over the long run, much more IT staff involvement than [SAN]," the argument being that it takes more resources to manage many small (their characterization) filers than fewer behemoth SANs. I think most on this list would refute that point for many reasons. 2. Without understanding the requirement you seek to address, it's hard to make pertinent comparisons between EMC's IP4700 and NetApp filers (I assume this is the comparison you're making, not Celerra vs NetApp). However, some general points to consider: - if you're doing anything other than home directories and general purpose file sharing, you should consider the level of application support for the two platforms (I believe you will find NetApp is more likely to have certified support on their filers, although in researching this I was surprised to find EMC had certified Oracle on IP4700; nonetheless, from my point of view EMC will want you to do anything other than home directories and file sharing on a Sym). - IP4700 RAID5 / NetApp RAID4 -- inherent dynamic RAID set expansion advantages to NetApp - IP4700 RAID set sizes are restricted to 5- or 10-disk multiples -- less user configurable - no competing offering to NetApp's SnapMirror from EMC for the IP4700 platform 3. Getting back to your specific question about admin... Generally speaking, EMC supports their own products. This support is surely world-class, but it also carries a world-class price tag. As an administrator you won't be adding storage capacity, loading patches or performing other typical tasks -- they will (unless you get and maintain expensive certification training). In today's bloody market (EMC lost over $1B last quarter) vendors are doing everything they can to move product. Getting demos should be easy from both (now that manufacturing has had time to catch up with the response to Sept 11 -- for a while all demos and stock from a number of h/w vendors were redirected to NYC and DC for disaster response). Get 'em both in your lab and try 'em out for yourself. Let us know what you find. Just my $.02. Joe > -----Original Message----- > From: Mark Johnson-Barbier [mailto:mjb@mj3.org] > Sent: Wednesday, October 17, 2001 1:02 PM > To: Toasters (E-mail) > Subject: Re: NetApp and EMC Comparisons > > > This is a good write-up about NAS vs. SAN in general. I got > the link from > EMC, but the recommendations seems to favor NetApp for the needs of my > company. > > http://www.clipper.com/bulletins/sanvnasfinal.pdf > > > > ----- Original Message ----- > From: "Killion, Brent (ZB7217)" <Brent.Killion(a)pinnaclewest.com> > To: "Toasters (E-mail)" <toasters(a)mathworks.com> > Sent: Wednesday, October 17, 2001 8:12 AM > Subject: NetApp and EMC Comparisons > > > > Fellow Filer Admins- > > > > Can anyone point me to any white papers comparing the EMC > NAS products > with > > Network Appliance NAS solutions? Specifically, I'm looking > for comparisons > > of ease of use and ease of administration. > > > > Thanks. > > > > >

1 0

RE: remove
by Braunschweiger, Udo 18 Oct '01

18 Oct '01

please send a mail to majordomo(a)mathworks.com <mailto:majordomo@mathworks.com> with the only word in the message body unsubscribe toasters that's all! -----Original Message----- From: Morgan, Joshua [mailto:JMorgan@Trivergent.com] Sent: Donnerstag, 18. Oktober 2001 17:02 To: toasters(a)mathworks.com Subject: remove remove Joshua Morgan PH: (864) 331-7939 jmorgan(a)trivergent.com

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