toasters

toasters@lists.teaparty.net

13515 discussions

Recovered error messages no longer logged in 5.1.2R3
by Greg Kitch 18 Feb '99

18 Feb '99

We recently upgraded our operating system version on an F220, F330 and F540 to 5.1.2R3. After a few days we noticed that we had stopped receiving recovered error messages on our SCSI disks. Following is an example of one such message: Sun Jan 17 02:19:16 PST [isp_main]: Disk 4a.4(0x4e55d0): READ sector 6200854 recovered error (1 17, 1) We sent a request to support inquiring about the status of these error messages and have been told that beginning with 5.1.2R3 they are no longer being logged. It is the case that the occasional and infrequent occurrence of these messages on a very active filer is not an indication of a problem, it is also our understanding that if these messages start increasing (1 an hour or 10-20 per day) on a specific disk for different blocks, then you should be paying very close attention to that disk and we would notify support. This seems to have been the case with several disk failures that we have experienced on our filers. An RFE (12820) has been filed at our request to supply a toggle for these messages to enable the customer to selectively include or exclude them from the messages file. We believe messages of this nature can be very helpful although one must understand their significance (or insignificance) when they occur. We have read some of the discussion concerning these messages on toasters but are not aware of anyone else that has noticed their disappearance and would like to see them made available again. Have I missed them? Does anyone else have an interest in having the flexibility of including soft error messages in the log files or is there a general consensus that they are unnecessary. We're trying to get support from other users for making the toggle available. One step further, we would like to see a list of all such error messages potentially logged by the system and what they mean. We have seen such lists on other systems and think that they could be helpful for the filers as well. Thanks. Greg Kitch UCLA SEASnet Computing Facility 2567 Boelter Hall Los Angeles, CA 90095

4 3

Netapp nfs over tcp
by Dave Heiland 18 Feb '99

18 Feb '99

I've currently got my filers with 'options nfs.tcp.enable off' from when I first set them up due to a perception that this would cause problems. Has anyone got tcp enabled? Have they seen any problems/benefits, or are there any known issues with this? We use mainly Solaris 2.5.1/2.7 machines and HPUX 10.20 machines, now with NFS v3 support added, doing compiles and standard (but heavy) file access (no news server, proxy, etc.) Dave

4 3

Recovered error messages no longer logged in 5.1.2R3
by Greg Kitch 18 Feb '99

18 Feb '99

1 0

Re: Poor small file CIFS performance in 5.2.1
by sirbruce＠ix.netcom.com 18 Feb '99

18 Feb '99

On 02/18/99 09:29:26 you wrote: > >Hi All, > >Any other CIFS users seeing very poor CIFS performance when accessing large >numbers of small files? > >The specific problem is with across network backups using ARCServe. We were >seeing 70MB/min to a DLT3 drive when running OnTap 5.1D3, but upgrding to >5.2.1 has dropped that back to 20MB/min. > >When I backup large files (256MB core files for example) I'm seeing backup >throughput of ~130MB/min which is pushing the DLT drive close to its limit >(150MB/min with 2:1 compression). > >In general the performance for large files is better in 5.2.1, but Netapp >seem to have broken the small file throughput. By large numbers of small files, do you mean a bunch of small files in one directory? If so, are you sure this collection of small files actually existed when you did the backup before on the previous OS version? I guess what I'm getting at is the possible explanation that nothing broke between 5.1D3 and 5.2.1. Obviously, if something did, you'll have to wait to hear back from Netapp. It's also possible there was a fix in your D3 patch that did not make it into 5.2.1. People running patched versions to address a particular problem should always check to see if the patch made it into a subsequent release. Bruce

1 0

Poor small file CIFS performance in 5.2.1
by Garrett Burke 18 Feb '99

18 Feb '99

Hi All, Any other CIFS users seeing very poor CIFS performance when accessing large numbers of small files? The specific problem is with across network backups using ARCServe. We were seeing 70MB/min to a DLT3 drive when running OnTap 5.1D3, but upgrding to 5.2.1 has dropped that back to 20MB/min. When I backup large files (256MB core files for example) I'm seeing backup throughput of ~130MB/min which is pushing the DLT drive close to its limit (150MB/min with 2:1 compression). In general the performance for large files is better in 5.2.1, but Netapp seem to have broken the small file throughput. I've been through the usual tests: check duplex, swap network cables, even moved both backup host and the filer to another switch. Given the throughput on large files, I can't see it being a network problem. Neither the switch nor the filer interface are showing large number of errors. Backup performance from other NT servers hasn't dropped off. We backup from a snapshot. The small files (source files mainly, we're a SW house) are in quota trees using mixed security. There's quite a complex ACL system in place and some files may have 6-8 NT groups as part of the ACL. cifs.oplocks.enable is on and cifs.lvl2_oplocks_cap is off, which hasn't changed since the upgrade. I have a support call open with Netapp (Log ID: 52724.00000) and have provided them with a 60s network trace of the traffic between the filer and backup host when the backup throughput has gone way down and there's not much other filer usage, but they have yet to get back to me. Can anyone think of anything I should check for? Thanks, GB

1 0

Re: Sharing files between NT and Unix
by Mark Muhlestein 17 Feb '99

17 Feb '99

Actually it depends on the application. If the Windows application has the file open exclusively, then you won't be able access it from NFS. But if it is open in a mode that allows readers, it will work. And some applications open a file, copy it to a temporary file, close the original file, operate on the copy, then rename it to the original name when you do a save. And byte range locking also comes into play with some applications. All these types of lock are supported in a multi-protocol manner on the filer. Mark Muhlestein -- mmm(a)netapp.com -----Original Message----- From: Tom Limoncelli <tal(a)research.bell-labs.com> To: Schepers, Jan <Jan.Schepers(a)nl.origin-it.com> Cc: 'toasters(a)mathworks.com' <toasters(a)mathworks.com> Date: Tuesday, February 16, 1999 7:32 PM Subject: Re: Sharing files between NT and Unix >"Schepers, Jan" wrote: >> "NetApps Filers can't share files between NT and Unix *simultaneously*." > >I do it every day! > >It should be noted that the exact same file can't be opened at the exact >same time (ie. you can't "cat" a file in unix at the exact moment >MS-Word is reading the file). This is because MS-Word locks the file! >So, if you find a product that WILl "cat" the file while MS-Word is >using it, it sounds like they aren't implementing any locking! Ouch! > >--tal >

1 0

cont: Sharing files between NT and Unix
by Schepers, Jan 17 Feb '99

17 Feb '99

>Hello People, > >This was my question to you: >Apparently I have never understood the concept of an netapp filer, >but this is what an HP guy is trying to tell me: >"NetApps Filers can't share files between NT and Unix *simultaneously*." > >What could he have meant ? Thanks to all the numerous people that gave their reaction. I realized that dropping the question amongst equal tuned people would result in responses that line with my observations. Ergo, the place the get the answer is of course the HP sales-rep. I'll let you know when it comes to something beautiful :-) MNY THKS, Jan >

1 0

Sharing files between NT and Unix
by Schepers, Jan 17 Feb '99

17 Feb '99

Hello People, Apparently I have never understood the concept of an netapp filer, but this is what an HP guy is trying to tell me: "NetApps Filers can't share files between NT and Unix *simultaneously*." What could he have meant ? cheers, Jan

5 4

Question on 5.x
by Jaye Mathisen 17 Feb '99

17 Feb '99

I'm still running 4.3.4, going to upgrade to 5.x shortly. One question: On the volumes that can be created, can I set up some to have snapshots, and some that don't? Or is it only system-wide?

2 1

Re: Netapp nfs over tcp
by Gordon Keegan 16 Feb '99

16 Feb '99

> On Fri, 12 Feb 1999, Gordon Keegan wrote: > > > > > Speaking of NFS over TCP, we'd also like to try turning nfs.tcp.enable > > on to do some testing. Does anyone know if this would affect current > > mounts if we just enable it at the filer command line with > > "options nfs.tcp.enable on", or can that be done? > > Your clients aren't currently mounting with tcp, you'll have to have > them umount the server and then remount after you've enabled it on > the server. So it is kind of a chore. > I probably didn't phrase the question as well as I should have. If I enable NFS over TCP, then mount a single host using TCP and do testing, will that stomp on the dozens of other clients that still have their UDP mounts? (there; hopefully that was better :) -- Gordon Keegan, M10-10 Phone: (508) 261-4696 Motorola ISG Fax: (508) 261-5757 20 Cabot Boulevard Mansfield, MA 02048 email: lgk011(a)dma.isg.mot.com

6 5

Re: Sharing files between NT and Unix
by mds＠gbnet.net 16 Feb '99

16 Feb '99

On Tue 16 Feb, 1999, "Schepers, Jan" <Jan.Schepers(a)nl.origin-it.com> wrote: > Hello People, > > > Apparently I have never understood the concept of an netapp filer, > but this is what an HP guy is trying to tell me: > "NetApps Filers can't share files between NT and Unix *simultaneously*." Well as the above statment is totally, provably incorrect... > > What could he have meant ? ...he evidently meant 'I will lie outrageously to try to boost my company's chances of a sale'. So many companies' sales-forces resort to this that I find myself proposing a cull of salespeople - in the Canadian-baby-seal style, to preserve the viability of the rest of the species. > > cheers, You're most welcome. > > Jan > >-- End of excerpt from "Schepers, Jan"

1 0

Re: Netapp nfs over tcp
by Daniel Quinlan 16 Feb '99

16 Feb '99

Actually, I have one question: is there any reason we wouldn't want to start using NFSv3/UDP for Solaris boxes on our switched LAN? - Dan

2 1

Shared Snapshot directory
by Patrick.Day.B＠bayer.com 16 Feb '99

16 Feb '99

This is kinda long, so i appologize ahead of time... We are running OnTap version 5.2.1 on a 740. When we ordered the system we got an autoloader backup (powerstor L500). Unfortuantly, Ontap can't handle the loader to meet our needs. So, as a solution, we attached the device to an NT server running BackupExec 7.2. I was thinking that in order to get a good backup of the system i could share out the .snapshot/nightly.0 as shapc$. This would give me a look at the system from the the root directory. At first it appeared that this would work fine, the CIFS SHARE command shows the snapc$ share as always being mapped to the nightly.0 directory, even after the nightly.0 is changed to nightly.1. However, when i did a snap list command, it shows that the nightly.1 is "busy". When i do a backup, it backs up the OLD snapshot directory and when i open the directory, i confirm that it is the old data. It seems that the Netfiler remembers which directory is mapped regardless of the name. I have found that when i reboot, the share is recreated to the correct directory, but rebooting every night is not an option. I have also changed the saved nightly snapshots to 1. The system will not create a new snapshot because the old one is busy and of course, having only one nightly snapshot is not an option. Is there any way to get the netfiler to refresh the share automatically? ---or make it stupid and not remeber that today's nightly.1 is yesterday's nightly.0? Any input would be appreciated. --Patrick Day Sys. Admin Bayer Corporation

2 1

Linux NFS over TCP
by Daniel Quinlan 16 Feb '99

16 Feb '99

FYI - with the current Linux release (version 2.2.1), NFS over TCP is now working (it went into 2.2.0-pre6). But, only NFS version 2. Anyone have an opinion on NFSv2 over TCP compared to NFSv2 over UDP? I seem to remember seeing some benchmark that showed it was better, but now I can only find benchmarks (like SPEC SFS 2.0) that show it has a faster response time, but not throughput. - Dan

1 0

Tape Drives Question????
by Jason Middlebrooks 15 Feb '99

15 Feb '99

Will a Redwood Tape Drive work with a Netapp Filer. Has anybody tried this yet??? Jason Middlebrooks Systems Engineer Datalink Phone: 888-933-9327x260 2264 Northwest Parkway, Suite I Pager: 888-502-1625 Marietta, GA 30067 Fax: 770-933-9272

1 0

Re: Netapp nfs over tcp
by mds＠gbnet.net 15 Feb '99

15 Feb '99

Oooooooooh. An idea is aborning.... How about allowing filers admins to specify the protocol capability profile that a client is allowed to see on a client-by-client basis in a file not unlike the name-mapping files and exports files and quotas files that we already use... Then we could have NFS-over-TCP turned on for WAN clients but turned off for LAN clients and can experiment with NFS3/2 - all by configuring at the server end (one place) rather than at the client end (many places). Thoughts?

3 2

subscription
by Michael Chong 13 Feb '99

13 Feb '99

Michael M. Chong Web Server Düde The Motley Fool http://www.motleyfool.com

1 0

RE: Can Netapp F700 series do load balancing via NT without using a Virtual IP?
by Sam Schorr 13 Feb '99

13 Feb '99

Keith, We have struggled with this same problem. However, in our case, the "clients" are actually IIS systems in a Virtual IP site. We had tried to have 2 NICs in the IIS NT "clients", one connecting to the shared net that also had the PDC, WINS, DNS and other servers on one subnet, while we had the filer and our SQL server on another sunet. The second subnet I'll call a "backchannel", the idea being to have requests from the IIS "clients" routed to the filer or database through the "backchannel". What we ran into was that even though we put this backchannel on its own switch, and even though we specifically called the filer by the IP address via UNC (ie, \\filerIP\sharename), NT still treats this as a string, not an IP. Thus, NT would try to resolve this through WINS, which it routed through the other subnet!! (the one the WINS server could 'see'). We tried many tricks including using the NOWINS and NOBROADCAST settings on the second NIC in the filer. We still have not been able to get a second subnet going. Can you suggest what we may be missing? Sam -----Original Message----- From: Keith Brown [mailto:keith@netapp.com] Sent: Friday, February 12, 1999 2:53 PM To: Mike Ball; toasters(a)mathworks.com Subject: Re: Can Netapp F700 series do load balancing via NT without using a Virtual IP? >I am assuming [Windows NT] sees the filer as a multihomed server and sees all four IP's >But does it load balance between the 4 ports on the quad card? A commonly asked question. I think I used to have a pointer to a Microsoft document that described this, but apparently that has gone walk-about, so I'm afraid you're stuck with my from-memory verbage below for now... The Windows NT client usually starts out with the filer's name. It uses this name to queery WINS and gets returned a list of IP addresses (assuming the filer is multihomed). It then goes through the list of returned addresses and compares them with the address on its own, usually single interface. First it looks for an address to use that is on the same IP subnet as it finds itself on. If it finds multiple of those, it picks one at random, which will tend to balance the load to a single filer that has multiple interfaces connected to that same subnet as the client. If it doesn't find a match, it will look for an interface in the same IP network as it finds itself on. Again, if it finds multiple of those it will pick one at random, tending to balance the load to the target's interfaces. Finally, if it finds none of the above, it will just pick an address to use at random and go for it. Per our 5.1 releases of Data ONTAP and higher, you also have the ability to control which interfaces the filer will register with WINS. This gives you the ability to stop certain Windows clients from choosing interfaces on the filer which they might not be able to access for various topological and/or routing issues. Keith

1 0

RE: FW: FW: URGENT!!!! FW: NetApp Filer software versions 5.x: po ten tial har dware killer (fwd)
by Madison, Shannon 12 Feb '99

12 Feb '99

"Jason Downs" wrote: > Oh, and by the way: Keep in mind that _I am a customer_. > Attacking me or making snide comments instead of > addressing the issues that I've raised does NOT encourage > me to purchase more Network Appliance products. Please accept my humblest apologies. There was never any intention on the part of myself, Radek, or Network Appliance to insult you or your opinions. -Shannon Madison

1 0

Re: Can Netapp F700 series do load balancing via NT without using a Virtual IP?
by Keith Brown 12 Feb '99

12 Feb '99

>I am assuming [Windows NT] sees the filer as a multihomed server and sees all four IP's >But does it load balance between the 4 ports on the quad card? A commonly asked question. I think I used to have a pointer to a Microsoft document that described this, but apparently that has gone walk-about, so I'm afraid you're stuck with my from-memory verbage below for now... The Windows NT client usually starts out with the filer's name. It uses this name to queery WINS and gets returned a list of IP addresses (assuming the filer is multihomed). It then goes through the list of returned addresses and compares them with the address on its own, usually single interface. First it looks for an address to use that is on the same IP subnet as it finds itself on. If it finds multiple of those, it picks one at random, which will tend to balance the load to a single filer that has multiple interfaces connected to that same subnet as the client. If it doesn't find a match, it will look for an interface in the same IP network as it finds itself on. Again, if it finds multiple of those it will pick one at random, tending to balance the load to the target's interfaces. Finally, if it finds none of the above, it will just pick an address to use at random and go for it. Per our 5.1 releases of Data ONTAP and higher, you also have the ability to control which interfaces the filer will register with WINS. This gives you the ability to stop certain Windows clients from choosing interfaces on the filer which they might not be able to access for various topological and/or routing issues. Keith

1 0

RE: Netapp nfs over tcp
by Thompson, Arnold 12 Feb '99

12 Feb '99

The change is dynamic, but be advised you will need to re-mount the clients to actually use NFS over TCP. > -----Original Message----- > From: Keith Brown [mailto:keith@netapp.com] > Sent: Friday, February 12, 1999 2:20 PM > To: toasters(a)mathworks.com; Gordon Keegan > Subject: Re: Netapp nfs over tcp > > > >Does anyone know if this would affect current > >mounts if we just enable it at the filer command line with > >"options nfs.tcp.enable on", or can that be done? > >Is it a fairly transparent thing or should we schedule > >down time? > > Down time? Heavens no! The words "down" and "time" are > considered very ugly > around these parts when used adjacently. :-) > > I had no reason to believe that this would not be the case, > but just to be on > the safe side I did a little double-check on a filer here for > you, and sure > enough you have nothing to be afraid of. Go ahead and enable > TCP at your > leisure... > > Keith > >

1 0

Re: Netapp nfs over tcp
by Keith Brown 12 Feb '99

12 Feb '99

>Does anyone know if this would affect current >mounts if we just enable it at the filer command line with >"options nfs.tcp.enable on", or can that be done? >Is it a fairly transparent thing or should we schedule >down time? Down time? Heavens no! The words "down" and "time" are considered very ugly around these parts when used adjacently. :-) I had no reason to believe that this would not be the case, but just to be on the safe side I did a little double-check on a filer here for you, and sure enough you have nothing to be afraid of. Go ahead and enable TCP at your leisure... Keith

1 0

Re: Bugtraq item about Netapps.
by sirbruce＠ix.netcom.com 12 Feb '99

12 Feb '99

On 02/11/99 20:22:28 you wrote: > >http://www.geek-girl.com/bugtraq/1999_1/0594.html > >I think that falls into the "Doc, it hurts when I do that" category, but >it's a good reminder anyway. The message, however, is just wrong in many places. I'll quote it below to pick apart. Jason Downs writes: >Filer's typically have an "admin host" which can mount and read/write to the >filer root directory. Without it, it's impossible to do any sort of system >maintenance on the filer. This simply isn't true; there's many sorts of system maintenance that can be done on the console without an admin host, and even moreso now with Web-based administation. One doesn't even have to have a permanent admin host... you could just briefly export the root directory for a quick update, then unexport it from the filer console. >If this host is compromised it's obviously bad news for the filer. But now, >apparently new with the 5.x revisions of the filer operating system, a >malicious individual can likely destroy the disk drive hardware itself. >It is not known if any sort of sanity check is done on the contents of the >firmware files; it's likely there is none, considering the type of code they >contain. This isn't new; a malicious individual could potentially effect firmware in previous versions. This is potentially the case in almost any OS... although I admit, 5.x makes it a little "easier" to do so. Firmware also isn't hardware, although bad firmware could theoretically lead to physical damage of the disk drive hardware mechanism. >Of course, it is trivial to gain command line access to a filer once the >admin host is compromised. They use what amounts to /etc/hosts.equiv for rsh >access. Wrong. People keep thinking the admin host is some mythical authoritative host. It isn't. It's nothing. Forget the term. You *can*, if you like, allow one or more hosts to telnet into the filer, rsh into it without a password, or mount it's root partitions. These are no more or no less a factor in the filer than in any other system, and you are perfectly capable of *not* allowing a host to do any of the above. The filer will continue to work. >It has always been important to make sure the "admin host" of a filer is >secure. This is true. >Now it seems Network Appliance has just raised the stakes; not >only can you lose your data, but you can also potentially lose hundreds >of thousands of dollars worth of hardware. This isn't true, and no one should be doing risk-analysis assuming that a user accessing a system through software can't do damange to the hardware underneath. Jason, I'm CC:ing you on this; you're free to insert it into the bugtraq archive record if you wish since I don't subscribe to it. Bruce

8 8

Can Netapp F700 series do load balancing via NT without using a Virtual IP?
by Mike Ball 12 Feb '99

12 Feb '99

I am trying to find out if you can do load balancing on a filer box w/o using a virtual IP? For example, if I have a quad Ethernet card installed in a F760 with the ports assigned to IP's 10.0.0.1, 10.0.0.2, 10.0.0.3 and 10.0.0.4 how does NT 4.0 deal with accessing the filer? I am assuming it sees the filer as a multihomed server and sees all four IP's. But does it load balance between the 4 ports on the quad card?

1 0

Can Netapp F700 series do load balancing via NT without using a Virtual IP?
by Mike Ball 12 Feb '99

12 Feb '99

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