toasters

toasters@lists.teaparty.net

13515 discussions

reducing the size of a NetApp -- temporarily
by Daniel Quinlan 10 Oct '97

10 Oct '97

We recently purchased an F630 system with 26 9GB drives (two shelf pairs of 13 drives). Because multiple RAID groups (and to a less extent, multiple filesystems) are not supported, I have concerns about bringing a 200GB filesystem into the world. All of the space won't be needed for at least several months. The main problems brought on by large filesystems and are backup dataset size (and time), but the biggest concern is that when a disk goes, the raid reconstruction affects more of your data if you have a single RAID group. (As well as worse fault tolerance, etc.) And since we can add new drives by swapping them in, it doesn't seem like it would hurt to only bring 13 drives online at first. Since we have an active filesystem, can someone tell me what the best route to do this is? Any caveats? I have had one report from someone at NetApp that I should erase disk labels (I don't know if he meant all of them, just the ones being removed, or the ones staying) before starting, and another that it would be fine to just yank the disks I don't want and newfs the remaining ones. Dan

2 1

subscribe gsterlin@jjprayc1.jnj.com
by Greg Sterling 09 Oct '97

09 Oct '97

subscribe gsterlin(a)jjprayc1.jnj.com

1 0

Re: raid swap question
by sirbruce＠ix.netcom.com 08 Oct '97

08 Oct '97

On 10/07/97 21:12:31 you wrote: >Can you "raid swap" and add more than one spare disk at a time? No... you want to raid swap, shove a disk in, wait for it to detect, raid swap again, shove another disk in, etc. The fastest way to do lots of disks at once is to halt and power off the filer, shove the disks in, and then turn it back on. Bruce

2 1

raid swap question
by Daniel Quinlan 08 Oct '97

08 Oct '97

Can you "raid swap" and add more than one spare disk at a time? Dan

1 0

Re: DLT7000 streaming
by sirbruce＠ix.netcom.com 07 Oct '97

07 Oct '97

On 10/07/97 13:31:50 you wrote: >Maybe it's the wrong question, but I remember from other drive types >that once you drop into start-stop mode (or shuffling), the throughput >goes through the floor. Our SE thought that a 7000 being fed below >streaming speed might be actually slower than a 4000 being fed at the >same speed. Hmmm... I must admit ignorance in this area; I suppose it is possible, but it's not something I would believe off the cuff. >I agree that network backup is not great, but we're managing (just >about) to get it to work at the moment using amanda and exabytes. I'd >like to keep to one system, so we'll probably modify amanda to use the >DLT (either locally or on the filer). > >Has anyone any hard figures for backup speeds both locally on a filer >and remotely via 100Mbps ethernet and rmt or rsh? We don't have >anything that supports NDMP at the moment, but might consider Budtool, >though it doesn't support all the machines we have at the moment. Well, I have some numbers for FDDI.... rmt was abyssmal (500 K/sec), but that was several releases ago. A more recent release, uses a third host to rsh a piped dump and restore across a dedicated FDDI network, drove the tape drive much closer to it's limit, but I don't have exact numbers. 100Mbps ethernet isn't great but I would expect you'd see numbers much closer to local backup speed if you use rsh, but still not as fast as local. >Our main interest in the 7000 is capacity (single tape backup at the >moment), rather than speed. What if it takes 2 tapes? Are you going to get a 7000 stacker, or will you have to spend a lot of effort breaking the backup into pieces? At least a 4000 stacker will be able to keep up with your expansion. FYI, my experience with 7000 on the filer suggests that the capacity isn't as great as you might hope. A DLT-IV cartridge on a 4000 could often get 35-40GB on it (almost max 2:1 compression), but on a 7000 I could only get about 50GB, not the 60-70GB expected with max compression. It was not clear if this was a Netapp issue or a 7000 issue... perhaps they have resolved it? You would have to inquire and find out. >I'm told that they've now released a wide card supported on the F230 >in the latest release, so you can attach a 7000. Neato. I wish they'd update the web pages to actually reflect this. Bruce

1 0

Re: DLT7000 streaming
by Alan Judge 07 Oct '97

07 Oct '97

>>>>> On Tue, 07 Oct 1997 06:28:55 CDT, sirbruce(a)ix.netcom.com said: Bruce> Perhaps you are asking the wrong question? The 7000 isn't going to Bruce> be any slower than the 4000, I don't think, whether you keep it fed Bruce> or not. Maybe it's the wrong question, but I remember from other drive types that once you drop into start-stop mode (or shuffling), the throughput goes through the floor. Our SE thought that a 7000 being fed below streaming speed might be actually slower than a 4000 being fed at the same speed. I agree that network backup is not great, but we're managing (just about) to get it to work at the moment using amanda and exabytes. I'd like to keep to one system, so we'll probably modify amanda to use the DLT (either locally or on the filer). Has anyone any hard figures for backup speeds both locally on a filer and remotely via 100Mbps ethernet and rmt or rsh? We don't have anything that supports NDMP at the moment, but might consider Budtool, though it doesn't support all the machines we have at the moment. Our main interest in the 7000 is capacity (single tape backup at the moment), rather than speed. Bruce> PS - Is the 7000 even supported on the F230? I'm told that they've now released a wide card supported on the F230 in the latest release, so you can attach a 7000. -- Alan

1 0

Re: DLT7000 streaming
by sirbruce＠ix.netcom.com 07 Oct '97

07 Oct '97

On 10/07/97 10:38:50 you wrote: >Our SE raised an issue about streaming with the DLT7000 and how the >drive can slow down a lot if you don't keep it fed. Has anyone any >experience with either local or remote DLT7000s? Can a 230 keep one >streaming? Can you stream one over 100Mb ethernet, or at least keep >it going at a reasonable rate? Perhaps you are asking the wrong question? The 7000 isn't going to be any slower than the 4000, I don't think, whether you keep it fed or not. It may be that even local backup can't feed a 7000 fast enough under many "real world" circumstances. So if you're looking at 7000 for speed reasons, that's one thing; but then, I wouldn't be doing network backup at all in that case. My experience with rmt is that you won't get anything near what you want; if you are using NDMP, maybe the performance is substantially better. If you want fast backup, keep it local; if you're concerned about capacity, get a 7000, local or not. Bruce PS - Is the 7000 even supported on the F230? Unless they released a wide or wide differential card, it's SCSI-2 only. Or do they have a narrow version of the 7000? In either case, I would suspect that performance for such a peripheral to be severely reduced when compared to the wide version.

1 0

DLT7000 streaming
by Alan Judge 07 Oct '97

07 Oct '97

We've just purchased an F230 and are considering our backup options. We have been considering connecting a DLT7000 to a local host and using it for both backing up the filer and our local stuff (100BaseTX connect). The other possibility is to connect the drive to the filer and backup other stuff remotely (though that might confuse our backup software). Our SE raised an issue about streaming with the DLT7000 and how the drive can slow down a lot if you don't keep it fed. Has anyone any experience with either local or remote DLT7000s? Can a 230 keep one streaming? Can you stream one over 100Mb ethernet, or at least keep it going at a reasonable rate? Perhaps a 4000 stacker might be better, but the capacity of the 7000 would be nice. -- Alan Judge Phone: +353-1-6046901 Indigo Network Operations Centre Fax: +353-1-6046948

2 1

Re: Consistent crashing on quota initialization
by sirbruce＠ix.netcom.com 07 Oct '97

07 Oct '97

On 10/05/97 23:47:29 you wrote: > > Could someone try reproducing this problem? I know it pretty much >requires a Netapp that you don't mind the risk of crashing, but maybe >there are some new Netapp owners out there. ;-) It's a long-known problem that rebooting during quota initialization is a Bad Thing (tm). It might have been fixed in 4.1; I'm not sure. I'm sure support will give you the bug ID and let you know the current status of the bug, and how to get you out of your current jam. As the old doctor joke goes - don't do that! Bruce

2 3

Re: quota size limits (was: "quotas to track disk usage")
by sirbruce＠ix.netcom.com 06 Oct '97

06 Oct '97

On 10/05/97 21:27:29 you wrote: > >sirbruce <sirbruce(a)ix.netcom.com> writes: > >> There was a bug submitted... that's all I know. It should be >> trivial to implement a clamping option like df_2gb_limit, unless >> there is something really annoying in the quotad protocol that >> prevents this. > >Since I'm only wanting to use quotas for usage tracking, not hard or >soft limits, the solution is to turn on only part of the quota system, >the clients don't need to know. Aha! Yes, this is exactly the conclusion I came to a while back, and hit myself for not thinking of it before... just mount noquota, even though the filer keeps tracking usage. Sorry, I thought I had mentioned that in my original email some weeks back. Bruce

1 0

Re: quota size limits (was: "quotas to track disk usage")
by sirbruce＠ix.netcom.com 06 Oct '97

06 Oct '97

On 10/05/97 17:36:18 you wrote: > >On 09/05/97 01:40:58 Daniel Quinlan wrote: > >>> Given these parameters: >>> >>> - I would like to use quotas to track user disk space usage (but not >>> limit it) > >Bruce <sirbruce(a)ix.netcom.com> replied: > >> Easily done, but there are some caveats. >> >> [...] >> >> 2. Some clients can't deal with quotas great thatn 8GB, and >> there is currently no option on the filer to limit this (this >> is filed as a bug), so you'd have to restrict your users to 8GB. >> If you have any over that, you don't want them in the quota system. >> >> So, you'd want something like: >> >> # Quota Target type disk files >> # ------------- ----- ---- ----- >> /newhome tree >> * user@newhome 8G >> >> The first line might not even be necessary, but it's worth >> including just for reference. Hmm, unless it breaks the >> clients the same way. Maybe the 8G can be omitted from >> the user entry as well? I don't know, I haven't tried it. >> The line might also be ncessary for the file to be parsed >> properly. >> >> What Netapp really needs to do is a default quota report by >> uid/gid type of command on the filer, to avoid the need to >> set up quotas to track usage. > >Since we have users using more than 8GB, this is a major problem. Agreed. >What operating systems have the 8 GB limit? I have no idea. I think both SunOS 4.x and 5.x had problems, but I'm not sure. I would expect that newer OSes would be less likely to have the problem (it's similar to the 2GB disk/partition size issue). >Has there been any work >to address these issues? There was a bug submitted... that's all I know. It should be trivial to implement a clamping option like df_2gb_limit, unless there is something really annoying in the quotad protocol that prevents this. >Will it work without the "8GB" in the user >entry? I don't know. Bruce

2 1

Consistent crashing on quota initialization
by Brian Tao 06 Oct '97

06 Oct '97

Could someone try reproducing this problem? I know it pretty much requires a Netapp that you don't mind the risk of crashing, but maybe there are some new Netapp owners out there. ;-) I have two F210's both running OnTap 4.0.4N_970627_1000 (this is what came installed on them). I create some quota trees, turn on the quotas, and then reboot the filer while the quotas are being initialized. After it comes back up, quotas are off. If I try to turn them on, the filer crashes. I've been able to do this several times to both F210's. An excerpt from the syslog is included below (I've forwarded the whole thing to Netapp support). Sun Oct 5 23:32:12 EDT [rc]: saving 55M to /etc/crash/core.1-small ("wafl_nfs_lookup: inconsistent directory entry {x0 0 525472 103351382} <quota.db> in {x0 0 555944 103344539}.") Sun Oct 5 23:32:57 EDT [rc]: saving 256M to /etc/crash/core.1 ("wafl_nfs_lookup: inconsistent directory entry {x0 0 525472 103351382} <quota.db> in {x0 0 555944 103344539}.") Sun Oct 5 23:35:20 EDT [rc]: relog syslog Sun Oct 5 23:29:29 EDT [tn_login_0]: Login from host: adminhost Sun Oct 5 23:35:20 EDT [rc]: NetApp Release 4.0.4N_970627_1000 boot complete. -- Brian Tao (BT300, taob(a)netcom.ca) "Though this be madness, yet there is method in't"

1 0

Re: reducing the size of a NetApp -- temporarily
by sirbruce＠ix.netcom.com 06 Oct '97

06 Oct '97

On 10/05/97 18:00:36 you wrote: > >And since we can add new drives by swapping them in, it doesn't seem >like it would hurt to only bring 13 drives online at first. Since we >have an active filesystem, can someone tell me what the best route to do >this is? Any caveats? Don't forget to calculate the space you'll need for snapshots and so on too... also, the filesystem optimizes much better if you have lots of free space rather than always running the ragged edge. You don't have to add a whole shelf at once, so if 14 or 15 drives makes more sense for you, do that. Also, feel free to leave one in as a hot spare. >I have had one report from someone at NetApp that I should erase disk >labels (I don't know if he meant all of them, just the ones being >removed, or the ones staying) before starting, and another that it would >be fine to just yank the disks I don't want and newfs the remaining >ones. Yes, yank all the disks you don't want and boot off floppy, and then newfs the filesystem. That is the best way, and won't take very long at all on an F630. Stay away from other shortcut, quickie schemes that can have lots of hidden dangers. Bruce

1 0

quota size limits (was: "quotas to track disk usage")
by Daniel Quinlan 06 Oct '97

06 Oct '97

On 09/05/97 01:40:58 Daniel Quinlan wrote: >> Given these parameters: >> >> - I would like to use quotas to track user disk space usage (but not >> limit it) Bruce <sirbruce(a)ix.netcom.com> replied: > Easily done, but there are some caveats. > > [...] > > 2. Some clients can't deal with quotas great thatn 8GB, and > there is currently no option on the filer to limit this (this > is filed as a bug), so you'd have to restrict your users to 8GB. > If you have any over that, you don't want them in the quota system. > > So, you'd want something like: > > # Quota Target type disk files > # ------------- ----- ---- ----- > /newhome tree > * user@newhome 8G > > The first line might not even be necessary, but it's worth > including just for reference. Hmm, unless it breaks the > clients the same way. Maybe the 8G can be omitted from > the user entry as well? I don't know, I haven't tried it. > The line might also be ncessary for the file to be parsed > properly. > > What Netapp really needs to do is a default quota report by > uid/gid type of command on the filer, to avoid the need to > set up quotas to track usage. Since we have users using more than 8GB, this is a major problem. What operating systems have the 8 GB limit? Has there been any work to address these issues? Will it work without the "8GB" in the user entry? - Dan

1 0

Help
by Jim Grinnell 01 Oct '97

01 Oct '97

help

2 1

RE: Oracle on Netapps
by Bruce Clarke 01 Oct '97

01 Oct '97

On Wednesday, October 01, 1997 9:48 AM, Marc Nicholas [SMTP:marc@hippocampus.net] wrote: > My first question would be: when Sybase? :-) > > (I'm just plain greedy...) > > > -marc > --- > Marc Nicholas - Hippocampus OSD Inc. > 416 979 9000 - fax: 416 979 8223 - http://www.hippocampus.net > 125 John St. - Suite #100 - Toronto - Ontario - M5V 2E2 - CANADA > "Inter/Intra/Extra[net] consulting, services, hardware and software sales" There is no technical reason that we currently know of that would prevent any other database from making similar use of a filer for storing its datastores. Care still needs to be taken with configuration as you would any time you are splitting the files of a particular application across platforms, but putting all the datastores for a particular database instance on the filer for any given RDBMS should work fine. As has already been mentioned, we here at Network Appliance talk to our Sybase database over NFS with stores on the filer and have done so for quite some time without problems. It is rumored that Sybase pulled back from allowing NFS access to its data but I have not been able to confirm this yet. We can say (empiracally at least) that it works. The real answer to your question is that we have had some introductory discussions with Sybase and will continue to work with them to gain their certification. You can be of assistance in this by mentioning to your Sybase representatives of your interest in getting your Sybase stores onto the filer. This is precisely how the intial interest was generated at Oracle when a very large Oracle customer indicated that Oracle really should support storage on a filer. I believe that similar interest expressed to Sybase will generate similar support. At the very least, if you have a test Sybase environment, I would give it a try. You will not be able to operate in "raw" Sybase mode, of course, but should still see similar if not increased performance over the network. Bruce Clarke

1 0

RE: Oracle on Netapps
by sirbruce＠ix.netcom.com 01 Oct '97

01 Oct '97

On 10/01/97 00:12:58 you wrote: >> Any questions feel free to contact me. > >My first question would be: when Sybase? :-) Always? Even Netapp has run Sybase on the filers for some time. Or is there some more specific Sybase feature that doesn't work over NFS? Bruce

2 1

RE: Oracle on Netapps
by sirbruce＠ix.netcom.com 01 Oct '97

01 Oct '97

On 10/01/97 01:11:19 you wrote: >Yup...you can't use a Sybase native filesystem...which is something Sybase >themselves recommend. > >We've been talking with our local NetApp reps about doing some evaluations >of Sybase on a native filesystem versus Sybase running against an NFS >filesystem just to see what the performance difference is like. Ah... well, I don't think the Oracle deal is going to allow Oracle "native filesystem" (i.e. raw) either. It's just so that Oracle will now let their stuff run over NFS if it's the filer. Or is there more to it than that? Bruce

1 0

RE: Oracle on Netapps
by Bruce Clarke 01 Oct '97

01 Oct '97

I am the technical contact in NetApp Marketing responsible for the Oracle relationship. The software that officially supports Oracle will be the 4.3 release of OnTap. This software goes to external Beta test this week and should be generally available by the middle of October. The changes to support Oracle were small but very relevant and will be documented in an upcoming whitepaper describing the correct configurations for NFS, Oracle, and OnTap to ensure maximum performance and reliability when storing Oracle database files. This paper will be available in October. Many customers have chosen to already run Oracle without these specific changes and have been having no problems. So I do encourage anyone interested in storing their datastores on a filer to start their testing immediately and to upgrade to OnTap 4.3 before going into production. Any questions feel free to contact me. Bruce Clarke Technical Manager - Business Development Network Appliance, Inc. Email: bclarke(a)netapp.com Phone: 408-367-3156 On Tuesday, September 30, 1997 11:14 AM, Brian Tao [SMTP:taob@nbc.netcom.ca] wrote: > Anyone have some technical information regarding the recent > announcement between Netapp and Oracle? Is there a new OnTap (5.0?) > that needs to be uploaded on the filers, or is it something that can > be done with the 4.x OS? What's special about this Oraclized version > of OnTap? > > http://www.netapp.com/news/level3b/news_rel_970922.html > > -- > Brian Tao (BT300, taob(a)netcom.ca) > "Though this be madness, yet there is method in't"

2 1

Oracle on Netapps
by Brian Tao 28 Sep '97

28 Sep '97

Anyone have some technical information regarding the recent announcement between Netapp and Oracle? Is there a new OnTap (5.0?) that needs to be uploaded on the filers, or is it something that can be done with the 4.x OS? What's special about this Oraclized version of OnTap? http://www.netapp.com/news/level3b/news_rel_970922.html -- Brian Tao (BT300, taob(a)netcom.ca) "Though this be madness, yet there is method in't"

1 0

Null ethernet cable and/or private network
by David Drum 25 Sep '97

25 Sep '97

Hello everyone, Sitting here watching perfmeter report network collisions on the machine my NetApp is connected to is making me once again interested in exploring the possibility of setting up a null ethernet, full duplex 100 Mbps ethernet, or private switched ethernet segment connection. If anyone has done this I would appreciate someone setting me off in the right direction. Hardware and software (i.e. informing in.routed to use a particular interface) advice sought. net.karma offered. Regards, David K. Drum david(a)more.net -- It's hard to be bored when you're as stupid as a line. [1] Reality has a tendency to be so uncomfortably real. [2] You can only measure the size of your head from the inside. [3] When you proceed deliberately, mistakes don't cascade, they instruct. [4] [1] Vernor Vinge [2] Neil Peart [3] Larry Wall [4] Stewart Brand

2 4

Single vs. dual shelf performance on F230
by Brian Tao 25 Sep '97

25 Sep '97

If I had an F230 with, say, 6 drives, would I see any significant performance gain by placing 3 drives per shelf, rather than having all 6 drives on one shelf (and thus one controller)? This will be for heavily randomized file accesses (e.g., mail or news spool). -- Brian Tao (BT300, taob(a)netcom.ca) "Though this be madness, yet there is method in't"

3 4

Re: Single vs. dual shelf performance on F230
by sirbruce＠ix.netcom.com 24 Sep '97

24 Sep '97

On 09/24/97 19:00:24 you wrote: > > If I had an F230 with, say, 6 drives, would I see any significant >performance gain by placing 3 drives per shelf, rather than having all >6 drives on one shelf (and thus one controller)? This will be for >heavily randomized file accesses (e.g., mail or news spool). You might see a minor performance improvement. I think the last time this sort of thing was measured it was reported to be negligible, and didn't really appear until you had 4 or 5 drives on one controller. But this was with different controllers, different drives, etc. so it may no longer apply. It probably can't hurt, and would certainly be worth trying. (Especially with random access... I suppose with a nice layout with good striping, it might be preferable to read it all from one controller at once... not sure how these things are handled at the low level.) Bruce

1 0

CIFS and NIS experiences?
by Daniel Quinlan 24 Sep '97

24 Sep '97

We're going to be using CIFS on our filers and I would like to hear about anyone's experiences with CIFS especially using it under NIS. I tried briefly enabling CIFS about a month ago on one of our filers, using NIS, and it worked, but the filer started having problems only a few minutes later, so I disabled CIFS and NIS. The problems were pretty severe ... of the "did someone shut nfs off?" variety, but I didn't have much of a chance to investigate. We are running 4.1d. We're using Solaris boxes as NIS servers. For my next try at CIFS, I'm thinking about not using NIS and just using a text copy of the passwd and group files. Has anyone else tried and/or had problems with CIFS under NIS? -- Daniel Quinlan (at work) Linux, our last best hope for Unix quinlan(a)transmeta.com http://www.pathname.com/~quinlan/

2 1

request
by David Drum 23 Sep '97

23 Sep '97

Hello, Since I know NetApp people are listening... I'd like to be able to use backslash line-continuation in the /etc files, e.g. /patches/Solaris_2.5.1_x86_Patches -ro,access=allinger.apg.more.net:\ b5fan.apg.more.net This ought not be too difficult to support in a future release. Regards, David K. Drum david(a)more.net -- It's hard to be bored when you're as stupid as a line. [1] Reality has a tendency to be so uncomfortably real. [2] You can only measure the size of your head from the inside. [3] When you proceed deliberately, mistakes don't cascade, they instruct. [4] [1] Vernor Vinge [2] Neil Peart [3] Larry Wall [4] Stewart Brand

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