toasters

toasters@lists.teaparty.net

13531 discussions

Re: Quota's, Qtrees, Old_WAFL, NewWAFL, etc...
by grant＠netapp.com 06 Jun '98

06 Jun '98

Hi Thomas, Since you mention quota trees, I'd like to jump in and discuss the benefits of quota trees within 5.1 (shipping soon) and backup: My recommendation is that you should arrange your data in quota trees whenever possible when running 5.1 for these reasons: 1. You can have different security models running on a per quota tree basis (ntfs, mixed (multi-protocol) or unix). 2. Pass I & II of dump has been optimized for full quota tree (as well as full volume) backups. 3. dump has a new Q option that allows all non-quota tree data to be backed up. This allows you to frequently backup quota trees, but perhaps backup /etc and other miscellaneous data on an infrequent basis and you don't need to backup the whole volume to do so. Cheers, Grant

1 0

Re: Quota's, Qtrees, Old_WAFL, NewWAFL, etc...
by Keith Brown 06 Jun '98

06 Jun '98

> >I see that you can only use qtree's in the root... >??? - What is the real difference between quotas and qtrees? Filer qtrees are the entities on which you can slap tree quota restrictions in the /etc/quotas file. There are three types of quotas supported; User, group and tree. The former two are largely independent of qtree's, although they can be used in conjunction with them. The three can combine in relatively complicated ways and I'd recommend reading chapter 10 of the Systems Administration Guide for a full explanation. >??? - with 4.3.4 I can do non-root quota's for pure CIFS users right? If you mean quotas on individual Windows/CIFS users that are not associated with qtrees, then yes. Effectively, you would just define a regular user quota for each client. You could use group quotas and default quotas too (default quotas are just special cases of user quotas). >example: > >/cifs_home/<NT_username1> user 100M 20k >/cifs_home/<NT_username2> user 100M 20k >/cifs_home/<NT_username3> user 50M 10k Right. This would be fine. Just remember that the first field, the quota target field for a user quota is ultimately there to specify the UID of the user you are putting a quota on. If you are going to use a directory pathname in this field, the UID for the user quota comes from the owner of that directory. For this to work, the directory /cifs_home/<NT_username1> would have to have an ownership of whatever UID <NT_username1> was equated to via <NT_username1>'s entry in /etc/passwd (or NIS). >??? - These are just quotas on user directories right? No. User quotas are *not* specifically associated with the users home directory. User quotas expressed this way limit the users consumption of file and space resources on the root volume by default. If you want to limit the users space on another volume, you'd use the: /cifs_home/<NT_username1> user@/vol/foo 100M 20K syntax. >??? - I don't need quota trees for this type of individual CIFS user >quotas do I? Not if the above description is what you were looking for. > >The manual says that the user must have an entry in /etc/passwd. >??? - Is this true in a pure NT/CIFS file space? Yes. >??? - If so, Why? (no UID/GID?!) Because the user and group quota mechanism needs the files and directories created on the file systems to be owned by the UID's and GID's it's trying to impose its limits on for things to work. >??? - Will this change with 5.x? Not initially in 5.1 no. Sorry... >Lastly... >??? - Should I wait to impliment quotas until AFTER I go with 5.x or >will my 4.3.4 quotas and qtrees transition over ok to 5.x? Whatever system you come up with in 4.3.4 will transition just fine to 5.X as long as you keep doing things largely the same way. If you start creating multiple volumes in 5.X, then obviously some quota configurations may be effected by that. If you stop routinely mapping Windows/CIFS clients to valid UID's and GID's through /etc/passwd (which becomes perfectly feasible if you are using the NTFS compatible file system security in 5.1), then you will lose user and group quota-ing capability for those users, at least until we sort out some quota enhancements that will let you place quotas on users who are entirely defined by a Windows NT SID. However, tree quotas should keep on truckin' quite nicely, limiting even these kinds of users (the unmapped kind). Keith

1 0

Quota's, Qtrees, Old_WAFL, NewWAFL, etc...
by Weeks, Thomas 05 Jun '98

05 Jun '98

Hi all! Just some basic Q's... We're at OnTap 4.3.4 right now using the old WALF file system... I have implimented quota's yet and was wondering: I am planning on going to 5.x here soon (as soon as I can see how you all fair with it) along with it's full ACL's and implimentation of WALF... I am not currently using qtree's or quotas... I see that you can only use qtree's in the root... ??? - What is the real difference between quotas and qtrees? ??? - with 4.3.4 I can do non-root quota's for pure CIFS users right? example: /cifs_home/<NT_username1> user 100M 20k /cifs_home/<NT_username2> user 100M 20k /cifs_home/<NT_username3> user 50M 10k <etc...> (using PDC/BDC NT authentication for usernames) ??? - These are just quotas on user directories right? ??? - I don't need quota trees for this type of individual CIFS user quotas do I? The manual says that the user must have an entry in /etc/passwd. ??? - Is this true in a pure NT/CIFS file space? ??? - If so, Why? (no UID/GID?!) ??? - Will this change with 5.x? Lastly... ??? - Should I wait to impliment quotas until AFTER I go with 5.x or will my 4.3.4 quotas and qtrees transition over ok to 5.x? Please reply ASAP.. Thanx.. Thomas Weeks GTE SYSTEMS ENGINEER Testing-Integration Center "The Pit" Brooks AFB, DSN 240-5444

1 0

Backup strategy suggestions
by vance＠corp.theglobe.com 05 Jun '98

05 Jun '98

We have recently acquired an F230 which, once it is in production, host about 20GB of data. At the rate which new data is being added, we expect this to hit 80GB within 6 months. At that point, we expect to divide the load between a couple filers to maximize throughput. We will have a daily 8 hour window in which to perform backups. The average file size is about 20kB, and files are never modified or renamed, only created or deleted. Given this, does anyone have any words of wisdom about how to tackle backups? Thanks in advance for any advice you might have- Vance ......................... ........... vance huntley .. director of technology ............ theglobe.com . vance(a)corp.theglobe.com .........................

1 0

Soft Quotas
by Arvind K Aggarwal 05 Jun '98

05 Jun '98

Hi, Is there any way to set soft quotas? What I mean by soft quotas: If I give a user a quota of 100MB and he/she exceeds it, I don't want the system to die on him. Based on how much the excess disk usage is, I would a) Like to send him a message b) Give him/her a warning and then c) Disallow further disk usage. This is some what like the "GOD" program on the old VAX systems, if someone remebers those things :-); I have heard that it was pretty good. Thanks in advance Arvind Aggarwal

1 0

subscribe
by Elena Samsonova 05 Jun '98

05 Jun '98

1 0

Replication
by vance＠corp.theglobe.com 29 May '98

29 May '98

We would like to use some form of replication to mirror our NetApp 230 to another NetApp in a remote location. Has anyone on the list tried to do this before? Any suggestions on how to tackle this problem? Thanks in advance for any help you can provide- Vance BTW - If it matters at all, we are looking to replicate ~2M files totaling about 50GB. The WAN link would be fairly fast (> 45Mb). ......................... ........... Vance Huntley .. Director of Technology ............ TheGlobe.com . vance(a)corp.theglobe.com . http://www.theglobe.com .........................

3 4

subscribe
by David B. Ritch 28 May '98

28 May '98

subscribe toasters

1 0

User Quota Checking
by Mike Mueller 25 May '98

25 May '98

We've just turned on user quotas and I am trying to figure out a way to allow users to check their current disk usage and quota limit. Since the filer is running rquotad, this would seem to be the way to go. But alas, the filer rquotad apparently wants quota [path] instead of quota [uid], just like its command line quota report. It returns "NO QUOTA" when queried via a client's quota command. Has anyone figured out a way to allow a general user to check their disk usage and quota limits? It looks like there may be a way through the MIB, but how could this be easily made into a command line script that a user could run? Better yet, is there a way to allow the user to interact with the rquotad deamon that is already running on the filer to get their quota report? If not, why is it running? Thanks for your insight! -- Mike

3 3

command reference guide discontinued in 5.x
by Tom Yates 22 May '98

22 May '98

here's a little someting to take our minds off how nice NTP support would be... would anyone from netapp care to comment on the suggestion that the command reference guide has been discontinued in 5.x? would anyone else care to comment on whether they use the command reference guide? perhaps some might care to say that it's about twice as useful as the system administrators' guide, despite weighing half as much? i certainly think so. i'd even pay for it. i'd think it was a bit rich, given the price of the toaster, but i'd shell out folding money for it. it's just too useful. any chance we could have it back? Tom Yates - Unix Chap - The Mathworks, Inc. - +1 (508) 647 7561 MAG#65061 DoD#0135 AMA#461546 1024/CFDFDE39 0C E7 46 60 BB 96 87 05 04 BD FB F8 BB 20 C1 8C

2 1

Re: don't run an F630 with the case open
by Mark D Simmons 21 May '98

21 May '98

Credit to our SE's for pointing this one out to us when they installed our F630's. I can see why this would drive you up the wall trying to diagnose a fail-to-boot when no-one's pointed out the change.

2 1

Re: don't run an F630 with the case open
by Guy Lauterbach 21 May '98

21 May '98

At 08:28 AM 5/21/98 -0700, Christina Sanfilippo wrote: >Ouch! That's got to hurt. I am sure Rich was glad he could fix your problem. > >For the F200/F300 and F500 series filers these filers can run with the case >open. >As you found out the F630 can't. Although some NetApp filers *can* be operated with the case open, it is not a good idea to run any of these machines with the box open. This is because an open cabinet changes the airflow within the filer and often results in reduced airflow to some devices within the box, allowing them to overheat. And that means increased failures. Please, close them up! rgl

1 0

Re: don't run an F630 with the case open
by Christina Sanfilippo 21 May '98

21 May '98

Ouch! That's got to hurt. I am sure Rich was glad he could fix your problem. For the F200/F300 and F500 series filers these filers can run with the case open. As you found out the F630 can't. Cheers, Christina At 08:27 AM 5/21/98 -0400, Tom Yates wrote: >d'oh of the day contest entry, just fyi: > >you can run an F330 with the case open, or at least ajar. i often do >after i've done work, before i seal the case down tight, to make sure that >it boots and i'm not going to have to undo everything to fix it. well, >don't try the same trick with an F630; it doesn't work. thanks to rich >ciammaichella at netapp for not roasting me when we discovered that this >was a major part of my F630 booting problem. > >for those who haven't looked closely inside an F630, there are no longer >cables going from the board to the front-of-case displays; there are now >fixed connectors that seat when the board tray is slid fully home. this >may be true for other high-end models, too - i don't know. > > Tom Yates - Unix Chap - The Mathworks, Inc. - +1 (508) 647 7561 > MAG#65061 DoD#0135 AMA#461546 > 1024/CFDFDE39 0C E7 46 60 BB 96 87 05 04 BD FB F8 BB 20 C1 8C > > > > > ++++++++++++++++++++****************** Christina Sanfilippo Customer Support Representative Network Appliance 2730 (Shipping:2820) San Tomas Expressway Santa Clara, CA 95051 (408) 367-3747 chrissan(a)netapp.com ============>>>>>>>>>>>>>> Tech Support Hotline:1-888-4-NETAPP (Remember to reference your log # if you have one.) _____________________________________________________ VISIT Netapp On the Web http://now.netapp.com

1 0

don't run an F630 with the case open
by Tom Yates 21 May '98

21 May '98

d'oh of the day contest entry, just fyi: you can run an F330 with the case open, or at least ajar. i often do after i've done work, before i seal the case down tight, to make sure that it boots and i'm not going to have to undo everything to fix it. well, don't try the same trick with an F630; it doesn't work. thanks to rich ciammaichella at netapp for not roasting me when we discovered that this was a major part of my F630 booting problem. for those who haven't looked closely inside an F630, there are no longer cables going from the board to the front-of-case displays; there are now fixed connectors that seat when the board tray is slid fully home. this may be true for other high-end models, too - i don't know. Tom Yates - Unix Chap - The Mathworks, Inc. - +1 (508) 647 7561 MAG#65061 DoD#0135 AMA#461546 1024/CFDFDE39 0C E7 46 60 BB 96 87 05 04 BD FB F8 BB 20 C1 8C

1 0

NTP or rdate.....
by gordon＠nabaus.com.au 20 May '98

20 May '98

Yes, put me on the voting list too. Netapp folk have been misteriously quiet regarding this thread. We know nearly ALL of the netapp dudes read from this list, so how about a response from Netapp? What is Netapps official position regarding NTP? That's my 2cents worth. The check is in the mail.... .... Gordon. Gordon Maudsley, Unix and Network dude, National Australia Bank.

1 0

Re: NetApp File Server (FS230) mounted directly on an NT server
by Keith Brown 20 May '98

20 May '98

Thomas... Taking your various issues one at a time... >It looks like while yes >there is indeed NT server authentication happening now, no matter WHO I >want to access the CIFS shares, they STILL have to MANUALLY PLACED into >the /etc/passwd file!!!! Right. That is the case in current Data ONTAP releases, and the root reason for this requirement is that the security model used by the WAFL file system today is a UNIX security model. In order to assess what a Windows user can and can not do to the files and directories resident on its file system, a filer has to associate each Windows connection (user) with a UNIX identity (a UID and GID(s)) so it can control the users access to data using the UNIX inode based security that it stores (read, write and execute for owning UID, owning GID and other). The /etc/passwd isn't really storing Windows users accounts per se, it's storing mappings between account names and UIDs/GIDs. >What is the POINT of full blown NT server >(domain) authentication if I STILL have to input >ALL of the users into the /etc/passwd file! The Windows clients still get to authenticate themselves in the normal way with the domain infrastructure on startup. The full Windows security/authentication protocols are enforced on the wire, and the users also wind up being subject to all the correct limitations and restrictions that the Windows administrator may have assigned to their account (times of day they are allowed to use the network and so forth...). >We have TONS of >UNIX users across campus who are on various NIS+ domains >and as a result many have overlapping UIDs! So called "overlapping UIDs" situations are not handled by the above mechanism today. If by overlapping UIDs you mean multiple users posessing the same UID, I thought this was generally regarded as a pretty bad idea even to start out with? Unless the accounts with identical UIDs have different lexigraphical user names, that's that. If they do, you could actually get to "break up" the different users with the same UIDs by allocating them different UIDs in the filers /etc/passwd file. However, if you configure the filer into your existing NIS infrastructure as a client, then the problem will remain. As an FYI, we have a utility available on our NOW web site that many of our customers have found useful. Check out CVTUSERS in the "Tools & Utilities" section of NOW. It will automatically generate passwd and group databases from a Windows NT domain infrastructure for you. >Is the ONLY way to get around this is by going to 5.x? Not a current 5.X (5.0) release, no. The resolution to this "problem" as described above is actually quite a bit larger than you might imagine, and is embodied in a major enhancement to the Data ONTAP microkernel and WAFL file system that is already announced and about to be unleashed shortly. Specifically, the solution lies in enhancing WAFL to maintain true Windows NT style security attributes (NTFS Security Descriptors) such that Windows client users can be treated as exactly what they are, Windows client users who might not even have UNIX accounts at all! :-) The new capability is briefly mentioned in the announcement of our previous Windows Network Software enhancement (WNS 2.0) at: http://www.netapp.com/news/level3b/news_rel_971103.html The new software is currently in external test, and has even been briefly reviewed by Windows NT Systems magazine in their article at: http://www.netapp.com/news/level3b/news_rel_971103.html (See the "Sidebar: Midyear Release" box toward the end of the article). I'll save all the details for another time, but suffice to say that what you have coming your way is not just a "resolution", it's a whole new solution in its own right. :-) >Is there any way to MOUNT a NetApp file server directly to an NT file >system and access it like unix does with a /mnt directory? Well, yes. But apparently you already know about it because you write: >I know about the menthod of mounting to NT via a drive letter... but >you can't re-share that as a resource under that the NT file system CAN >you? You mean "re-sharing" a network drive or the directories thereon? No, you can't do that with current versions of Windows NT, at least not without resorting to some undocumented tricks which nobody has ever been able to confirm to me *really* exist. It's like re-exporting NFS mounted file systems on a UNIX machine, arguably not a good idea and definately not allowed. >You know.. like this: > given that > NetApp box="\\barium\C$" > > and > NT Server="\\ntbox" > >net use \\ntbox\shares\share \\barium\share >net use \\ntbox\shares\winapps \\barium\winapps What? Now there's a "net use" syntax that I've never seen before. >Anyone wanna tell me exactly why things can't be done like this? Because it would be a syntax error? :-) Keith

1 0

Re: command reference guide discontinued in 5.x
by applix!jimsparc!jim＠uunet.uu.net 19 May '98

19 May '98

>command reference guide has been discontinued in 5.x? you're kidding, right? You have to be kidding! -- Jim Morton, Applix Inc., Westboro, MA jim(a)applix.com

1 0

Re: Support for Network Time Protocol?
by mer＠world.evansville.net 19 May '98

19 May '98

Another vote for NTP support. It doesn't need to be full-fledged -- client-only is fine. The rsh/rdate thing is a kludge. -- Marc Rouleau VP and Chief Technology Officer Voice: (812) 479-1700 Fax: (812) 479-3439 Network WCS http://www.networkwcs.com

5 4

As long as we're voting
by Amy Chused 19 May '98

19 May '98

I'd vote for NTP too. Amy Chused Server Administration GTE Internworking

1 0

Re: Support for Network Time Protocol?
by Chris March : TTSG : 781-393-3236 19 May '98

19 May '98

I'll add my vote to get support for NTP. Chris March Thomson Technology Services Group

1 0

Re: Support for Network Time Protocol?
by Don Glascock 19 May '98

19 May '98

> Another vote for NTP support. ... And another: we too have asked for NTP support, but have not heard when this will occur. Our toaster has fulfilled its charter well ("a box that encapsulates file service for NFS & CIFS clients") except in this regard: a file's modification time can be critical in certain situations, and therefore must be accurate. Don Glascock -- Donald S. Glascock Mailstop GU-1016 Special Purpose Processor Development Group Mayo Foundation Rochester, MN 55905 Glascock.Donald(a)Mayo.EDU +1.507.284.1850 +1.507.284.9171 (fax) http://www.mayo.edu/sppdg/sppdg_home_page.html "No matter where you go, there you are."

1 0

Re: syslog and /etc/messages format
by Mark D Simmons 19 May '98

19 May '98

(after with much rude elision) Jason wrote: > We currently use logsurfer and swatch to parse syslog messages > generated by our F330. We also use procmail to forward "critical" > email from the filer to alpha-pagers. We use swatch too. > There isn't an appendix which contains a _complete_ > list of all error messages and sample _syntax_ in the filer > documentation. Tech Support couldn't even get me a list of all the > possible email subject headers the filer might send in case of error! You know that's something that IBM and Digital used to do religiously, and it's in most good books on the subject of Technical Documentation. One wee book or greppable ASCII file on-line that contains the full gamut of error messages (even if what the message means is sanitised to 'phone NetApp' for some of the really weird ones) would be great. > So, I'm all for more specific messages -- as long as they are > documented. Please. Ditto.

1 0

Re: Support for Network Time Protocol?
by Mark D Simmons 19 May '98

19 May '98

We also asked for NTP support a while back. I've currently got all 9 main filers rdating to an NTP'd Unix box every hour and so far everything is okey, but proper NTP support would reduce the niggly tasks on the admin server and my confidence in the time on the filers. 2p

1 0

Re: Support for Network Time Protocol?
by Daniel Quinlan 19 May '98

19 May '98

Jason D. Kelleher <kelleher(a)susq.com> writes: > I recall asking about NTP over two years ago. Very disappointing. I'm also interesting in NTP support, but I think I'd like to see the current software become a lot more reliable (and complete) before adding another new feature. All I need is more NetApp crashes because of NTP bugs. Dan

1 0

Re: Support for Network Time Protocol?
by sirbruce＠ix.netcom.com 19 May '98

19 May '98

On 05/18/98 09:14:26 you wrote: > >sirbruce(a)ix.netcom.com wrote: >> Okay, I was just curious. Most places I've seen aren't as "religious" about >> the use of NTP. Not that there's anything wrong with that. :) >> >> Bruce > >NTP is integral to keeping the NYSE, SIAC, and Amex clocks in sync. We >use it on everything so that all trading is accurate on all systems to >within milliseconds of each other, which is crucial when billions of >messages have to be ordered, stored, processed, and then compared at the >end of the trading day. I can imagine! The demands there are much different from a college campus or a departmental server in a small corporation. Different environments need different solutions. I think all these posts are great - the more Netapp knows about customer desire for this, the more likely they are to implement the feature. It certainly seems more widespread than the last time I saw the question asked (although this mailing list doesn't exactly constitute a scientific survey). Bruce

1 0

Jump to page:

2026

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