toasters

toasters@lists.teaparty.net

3 participants
13514 discussions

RE: Q: lose shelf power supply
by Alvarado, Michael 07 Jun '99

07 Jun '99

One other point to emphasize is that power supplies are fairly reliable devices and a redundant supply can be purchased for even greater protection. -----Original Message----- From: hitz(a)netapp.com [mailto:hitz@netapp.com] Sent: Monday, June 07, 1999 10:51 AM To: george(a)YorkU.CA Cc: toasters(a)mathworks.com Subject: Re: Q: lose shelf power supply > Could someone please explain what happens when you loose a single > power supply in a shelf? > > Is the filer smart (dumb) enough to halt? > Do I get 'double disk error' and loose the filesystem? The filer does the "right thing:" If there's a double disk failure in a single RAID group, you have definitely lost the file system, at least for now, and the filer tells you so. However, it immediately stops writing to the filed file system, so that it doesn't screw things up in case you do find those failed disks. So in the case you are talking about, the filesystem will appear to fail, but if you replace the failed power supply, everything should be okay again. (I don't know whether or not this sequence causes/requires a reboot. My guess would be that it does.) Dave

1 0

Re: Q: lose shelf power supply
by sirbruce＠ix.netcom.com 07 Jun '99

07 Jun '99

On 06/07/99 09:34:03 you wrote: > >Could someone please explain what happens when you loose a single >power supply in a shelf? That depends on whether or not you have another pwoer supply. If you do, there's no problem. >Is the filer smart (dumb) enough to halt? >Do I get 'double disk error' and loose the filesystem? Both? Actually, I can't say for sure what recent version of the OS does, but in past versions it did think that a bunch of disks had stopped responding, leading to a 'double disk error' scenario. However, this does not mean you have lost the filesystem, because you didn't actually *lose* any disk in this scenario, and the disks aren't marked bad because they aren't "there" to be marked. There was a bug/enhancement request filed which suggested that the filer recognize when a whole shelf fails and provide a more sensible error message. There may have been a similar change made when multivolume and multiraidgroup support was added, since the filer will want to keep running even when it loses a particular volume and not halt. However, I am assuming from your question that you want to know what happens when there's only one filesystem and no failover filer. >There is a discussion going on here as to what happens in a hardware RAID >configuration vs a NetApp filer when a power supply fails. > >The thought is that if I have only one power supply in a shelf and it >fails (failing 7 disks) it is like having multiple disks fail and >therefore you loose the filesystem. Like I said, in the past it certainly did "think" it had multiple disks fail, but this didn't make you lose the filesystem. If you tried to bring the filer back up sans 6 or 7 disks (and only one filesystem), it will fail to boot. Once you replace the power supply and turn it back on with all the disks attached, it will detect they are all there and pick up just fine. (You also have the option of just moving the disks to another shelf if you have empty slots, but I'm not sure if this works with fiber channel and/or when you have multiple raid groups.) There are a few possible problems that can still result. Data will not be lost, since it's all still in NVRAM, but this type of failure could result in your parity disk for the raid group being reconstructed, degrading performance for a while. This also sounds like a case where it's possible the disk labels can get confused and you'll wind up with disks on the failed shelf not matching the other disks, and thus the filer refusing to recognize them even after you replace the power supply. I am not saying this will happen (it shouldn't), but the exact circumstances when the labels can get out of sync is unclear to me. Anyway, this is also easily solved via a floppy command and won't result in any data loss. Finally, the loss of a power supply could potentially damage a disk and/or its contents, giving you a failed disk when you finally get back up or even data corruption that eventually necessitates doing a wack on the filesystem. However, these would be rare, extreme cases, generally the result of an external event killing the power supply (like a lightning strike or power surge) rather than the power supply failing on its own. But this type of problem isn't unique to Netapp; it could happen to any RAID system. The bottom line is that the vast majority of the time, once you replace the power supply the system will come back up just fine, and those times when there is a problem, the vast majority of them will be easily corrected without data loss. Bruce

1 0

Re: Whats better: several narrow or one wide
by sirbruce＠ix.netcom.com 06 Jun '99

06 Jun '99

On 06/03/99 13:44:32 you wrote: > >Hi Ya'll, > >This is probably one of those questions for which there's no >true correct answer, but I'm interested in peoples opinion. Okay. Just for the record, my opinion is based on a balance of administrative ease and flexibility with the need for high performance. >1) In setting up a NetApp, is it better to use one big/wide >connection (say gigabit ethernet) as opposed to a bunch of >narrow connections (say, 4 100mbps ethernet in a etherchannel). >(Note that for this discussion I'm ignoring the redundancy in >a etherchannel.) One big connection, say Gigabit, is ideal for your primary data backbone. You can still use other, smaller connections for other smaller networks. >2) On a Unix system that will be doing heavy NFS work to a NetApp, >is it better to use separate mounts on separate but smaller ethernet >connections (100mbps ethernet), or, run all mounts through one wide >ethernet connection (gigabit ethernet)? Again I would recommend running all the mounts from a particular machine to one filer through one particular large network connection. This is going to be administratively advantageous in the long run. Even on a network with lots of networks for data access with lots of different purposes (and I have no objection with such a configuration), I'd still recommend keeping all of the mounts from a given dataless client over the same network to the filer(s), unless there is some very specific reason not to (like dedicated remote backup networks). Bruce

3 2

Whats better: several narrow or one wide
by Richard L. Rhodes 05 Jun '99

05 Jun '99

Hi Ya'll, This is probably one of those questions for which there's no true correct answer, but I'm interested in peoples opinion. 1) In setting up a NetApp, is it better to use one big/wide connection (say gigabit ethernet) as opposed to a bunch of narrow connections (say, 4 100mbps ethernet in a etherchannel). (Note that for this discussion I'm ignoring the redundancy in a etherchannel.) 2) On a Unix system that will be doing heavy NFS work to a NetApp, is it better to use separate mounts on separate but smaller ethernet connections (100mbps ethernet), or, run all mounts through one wide ethernet connection (gigabit ethernet)? Specifically, I'm going to be setting up an Oracle DB on a NetApp. I'm looking at whether I would be better off using multiple 100mbps ethernets or one gigabit ethernet connection. Thanks in advance for all opinions!!!!!!!!! rick ---------------------------------------------------------------------- Richard L. Rhodes e: rhodesr(a)firstenergycorp.com Ohio Edison Co. p: 330-384-4904 f: 330-384-2514

2 2

Re: Whats better: several narrow or one wide
by Michael Salmon 04 Jun '99

04 Jun '99

+----- On Thu, 03 Jun 1999 13:44:32, "Richard L. Rhodes" writes: | Hi Ya'll, | | This is probably one of those questions for which there's no | true correct answer, but I'm interested in peoples opinion. | | 1) In setting up a NetApp, is it better to use one big/wide | connection (say gigabit ethernet) as opposed to a bunch of | narrow connections (say, 4 100mbps ethernet in a etherchannel). | (Note that for this discussion I'm ignoring the redundancy in | a etherchannel.) | | 2) On a Unix system that will be doing heavy NFS work to a NetApp, | is it better to use separate mounts on separate but smaller ethernet | connections (100mbps ethernet), or, run all mounts through one wide | ethernet connection (gigabit ethernet)? | | Specifically, I'm going to be setting up an Oracle DB on a NetApp. | I'm looking at whether I would be better off using multiple 100mbps | ethernets or one gigabit ethernet connection. In my experience using a single big whatever is best unless you know precisely what your are going to be doing with it. A single big whatever is much more flexible. Having said that though I am not sure that a single gigabit connection is going to give you more throughput that an ether-channel, your really comparing 2 sorts of big. The same thing applies on the Unix side, I would probably be more inclined towards ether-channel on both. One thing that I have found is that having large differences in the speed of your server and client nearly always causes problems. /Michael

1 0

RE: Restoring DOT 4.x tapes on 5.X?
by Melvin, Grant 04 Jun '99

04 Jun '99

Hi Benjy, We work very hard in the File System Recovery group to maintain both backwards & forwards compatibility with our dump format. We also strive to be compatible with Solaris ufsrestore. If you have any restore issues then we'd like to hear about them. The best mechanism for this is to open a call record with NetApp support. Cheers, Grant -----Original Message----- From: Benjy Feen [mailto:benjy@feen.com] Sent: Thursday, June 03, 1999 10:11 AM To: toasters(a)mathworks.com Subject: Re: Restoring DOT 4.x tapes on 5.X? On Thu, 3 Jun 1999, Jason D. Kelleher wrote: > Haven't yet had the need. I have however restored files from > both DOT 4.X & 5.X with ufsrestore on a Solaris 2.6 system. Is there a recipe to do this, or did it "just work" for you? We've been unsuccessful thus far in doing restores to Solaris 2.6... -=- Benjy Feen benjy(a)feen.com http://www.feen.com Specialization is for insects. =-=

1 0

Re: Restoring DOT 4.x tapes on 5.X?
by Jason D. Kelleher 03 Jun '99

03 Jun '99

In message <Pine.LNX.4.10.9906030409500.22731-100000(a)lucien.blight.com>, Ben jy Feen writes: >Anyone ever successfully restored a tape created with DOT 4.X on a 5.X >toaster? Haven't yet had the need. I have however restored files from both DOT 4.X & 5.X with ufsrestore on a Solaris 2.6 system. jason

2 1

None
by Jim Sanford 03 Jun '99

03 Jun '99

I've just downloaded amanda in an effort to help me with a problem on an HP. First off, I'm impressed with the product. However, my particular problem is that HPUX wont recognize the norewind tape device for fbackup (it's local backup utility) altho tar likes it just fine. My question then is, how does amanda use the HP norewind device? Am I correct in presuming it just dumps tars/gtar to the device? How does it handle incrementals? If I did a find . -mtime -1 | tar ..... I'd pick up all the files created/modified within the last day but some of these files would have the same names (eg .dt/... files) altho they'd reside in different directories. Unless the path was hardcoded to each, they'd overwrite each other on the tape, right? And if they path is hardcoded, how could tey be restored to different directories? So is amanda just a client/server package layered onto gtar? Thanks ********************************* Jim Sanford IS Department Research Systems, Inc. *********************************

3 2

Restoring DOT 4.x tapes on 5.X?
by Benjy Feen 03 Jun '99

03 Jun '99

Anyone ever successfully restored a tape created with DOT 4.X on a 5.X toaster? -=- Benjy Feen benjy(a)feen.com http://www.feen.com Specialization is for insects. =-=

1 0

Re: CIFS
by Matt Day 03 Jun '99

03 Jun '99

>Is or has anyone experienced any problems getting cifs to run on the netapps ? > We have an F720 and >are having performance problems with our win95 users. Just transfering files > from a server drive to the >F720 is very slow. The other problem is that several programs wont run if the > files have multiple access, >such as access. We changed the cifs.oplocks.enable to off .... Can anyone > shed some light ? Hi Mark, Sounds to me like you're probably running into a known CIFS bug. What version of Data ONTAP are you running? I strongly recommend contacting Netapp support about getting a software upgrade. Matt

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