toasters February 1999

toasters@lists.teaparty.net

106 participants
95 discussions

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

Re: FW: FW: URGENT!!!! FW: NetApp Filer software versions 5.x: poten tial har dware killer (fwd)
by Jason Downs 12 Feb '99

12 Feb '99

In message <7F608EC0BDE6D111B53A00805FA7F7DA02C3834B(a)TAHOE.netapp.com>, "Madison, Shannon" writes: >Forwarded per Radek: > >-----Original Message----- >From: Radek Aster >Sent: Friday, February 12, 1999 11:20 AM >Subject: Re: FW: URGENT!!!! FW: NetApp Filer software versions 5.x: >potential har dware killer (fwd) > > > >IMHO, this is a pile. Jason makes the statement that he can create a file >(of the appropriate size), fill it will garbage, and download it to disk >drives which will then become bricks. > >Geez. How stupid does he think Seagate is? Don't answer that. :-) It's not a question of how stupid I think Seagate is. It's a question of how stupid I think NetApp is for implementing 'online' disk firmware updates. >Seriously, the firmware files have checksums embedded in them. As part of >the update process, the drive will verify the checksum before committing >the firmware to flash. If the checksum doesn't verify, the update is >cancelled. No harm, no foul. Pretty SOP with firmware downloads. Heck, one >could make the same "security" argument with any hardware component with >downloadable firmware. Why pick on drives? All a checksum does is make it more difficult than dd'ing /dev/zero in order to produce something the drive will take. You're missing the point. >Granted, he *could* get his hands on unqualified and/or bad firmware and >download it to the drives .... is this enough to cry "the sky is falling"? > If this is seem as a serious enough "security issue", we can always ship >*encrypted* files, and decrypt them ourselves before downloading, thereby >verifying the contents and identity of files we ship. IMHO, the 'best' way to do it would be to require the files be on a floppy, or some other method which restricts firmware updates to the physical console. 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. >--Radek > >Jason Downs downsj(a)downsj.com <mailto:downsj@downsj.com> writes: > >Jason> I was going through the documentation for version 5.2.1 >Jason> (the latest) of the Network Appliance Filer operating system when I >Jason> stumbled upon this little gem: "Use the disk_fw_update command to >Jason> update out-of-date firmware on all disks or a specified disk on a >Jason> filer. Each filer is shipped with a /etc/disk_fw directory that >Jason> contains the latest firmware revisions." > >Jason> [...] > >Jason> "In the /etc/disk_fw directory, the firmware file name >is >Jason> in the form of product_ID.revision.LOD. For example, if the firmware >Jason> file is for Seagate disks with product ID ST19171FC and the firmware >Jason> revision is FB37, the file name is ST19171FC.FB37.LOD. The revision >Jason> in the file name is the number against which the filer compares each >Jason> disk's existing firmware revision." > >Jason> [...] > >Jason> "Before Data ONTAP 5.2, the disk_fw_update command copi >ed >Jason> firmware files from the /etc directory. In the /etc directory, the >Jason> name for the firmware file was in the form of product_ID.LOD. The >Jason> revision number was not included in the file name. Data ONTAP 5.2 >Jason> continues to support firmware files in the /etc directory for >Jason> backward compatibility. That is, if you obtain a disk firmware file >Jason> and store it in the /etc directory, you can use the disk_fw_update >Jason> command to copy that firmware file to disks, unless there is also a >Jason> firmware file for the same product ID in the /etc/disk_fw directory. >Jason> The files in the /etc/disk_fw directory take precedence over the >Jason> files in the /etc directory." > >Jason> [...] > > >Jason> Filer's typically have an "admin host" which can mount >and >Jason> read/write to the filer root directory. Without it, it's impossible >Jason> to do any sort of system maintenance on the filer. If this host is >Jason> compromised it's obviously bad news for the filer. But now, >Jason> apparently new with the 5.x revisions of the filer operating system, >Jason> a malicious individual can likely destroy the disk drive hardware >Jason> itself. It is not known if any sort of sanity check is done on the >Jason> contents of the firmware files; it's likely there is none, >Jason> considering the type of code they contain. Of course, it is trivial >Jason> to gain command line access to a filer once the admin host is >Jason> compromised. They use what amounts to /etc/hosts.equiv for rsh >Jason> access. It has always been important to make sure the "admin host" >Jason> of a filer is secure. Now it seems Network Appliance has just >Jason> raised the stakes; not only can you lose your data, but you can also >Jason> potentially lose hundreds of thousands of dollars worth of hardware. > >Jason> -- Jason Downs downsj(a)downsj.com <mailto:downsj@downsj. >com> -- Jason Downs downsj(a)downsj.com Little. Yellow. Secure. http://www.openbsd.org/ Sending unsolicited commercial email to this address may be a violation of the Washington State Consumer Protection Act, chapter 19.86 RCW.

1 0

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

12 Feb '99

Forwarded per Radek: -----Original Message----- From: Radek Aster Sent: Friday, February 12, 1999 11:20 AM Subject: Re: FW: URGENT!!!! FW: NetApp Filer software versions 5.x: potential har dware killer (fwd) IMHO, this is a pile. Jason makes the statement that he can create a file (of the appropriate size), fill it will garbage, and download it to disk drives which will then become bricks. Geez. How stupid does he think Seagate is? Don't answer that. :-) Seriously, the firmware files have checksums embedded in them. As part of the update process, the drive will verify the checksum before committing the firmware to flash. If the checksum doesn't verify, the update is cancelled. No harm, no foul. Pretty SOP with firmware downloads. Heck, one could make the same "security" argument with any hardware component with downloadable firmware. Why pick on drives? Granted, he *could* get his hands on unqualified and/or bad firmware and download it to the drives .... is this enough to cry "the sky is falling"? If this is seem as a serious enough "security issue", we can always ship *encrypted* files, and decrypt them ourselves before downloading, thereby verifying the contents and identity of files we ship. --Radek Jason Downs downsj(a)downsj.com <mailto:downsj@downsj.com> writes: Jason> I was going through the documentation for version 5.2.1 Jason> (the latest) of the Network Appliance Filer operating system when I Jason> stumbled upon this little gem: "Use the disk_fw_update command to Jason> update out-of-date firmware on all disks or a specified disk on a Jason> filer. Each filer is shipped with a /etc/disk_fw directory that Jason> contains the latest firmware revisions." Jason> [...] Jason> "In the /etc/disk_fw directory, the firmware file name is Jason> in the form of product_ID.revision.LOD. For example, if the firmware Jason> file is for Seagate disks with product ID ST19171FC and the firmware Jason> revision is FB37, the file name is ST19171FC.FB37.LOD. The revision Jason> in the file name is the number against which the filer compares each Jason> disk's existing firmware revision." Jason> [...] Jason> "Before Data ONTAP 5.2, the disk_fw_update command copied Jason> firmware files from the /etc directory. In the /etc directory, the Jason> name for the firmware file was in the form of product_ID.LOD. The Jason> revision number was not included in the file name. Data ONTAP 5.2 Jason> continues to support firmware files in the /etc directory for Jason> backward compatibility. That is, if you obtain a disk firmware file Jason> and store it in the /etc directory, you can use the disk_fw_update Jason> command to copy that firmware file to disks, unless there is also a Jason> firmware file for the same product ID in the /etc/disk_fw directory. Jason> The files in the /etc/disk_fw directory take precedence over the Jason> files in the /etc directory." Jason> [...] Jason> Filer's typically have an "admin host" which can mount and Jason> read/write to the filer root directory. Without it, it's impossible Jason> to do any sort of system maintenance on the filer. If this host is Jason> compromised it's obviously bad news for the filer. But now, Jason> apparently new with the 5.x revisions of the filer operating system, Jason> a malicious individual can likely destroy the disk drive hardware Jason> itself. It is not known if any sort of sanity check is done on the Jason> contents of the firmware files; it's likely there is none, Jason> considering the type of code they contain. Of course, it is trivial Jason> to gain command line access to a filer once the admin host is Jason> compromised. They use what amounts to /etc/hosts.equiv for rsh Jason> access. It has always been important to make sure the "admin host" Jason> of a filer is secure. Now it seems Network Appliance has just Jason> raised the stakes; not only can you lose your data, but you can also Jason> potentially lose hundreds of thousands of dollars worth of hardware. Jason> -- Jason Downs downsj(a)downsj.com <mailto:downsj@downsj.com>

1 0

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

12 Feb '99

On 02/11/99 23:28:15 you wrote: > >In message <199921122746541(a)ix.netcom.com>, > sirbruce(a)ix.netcom.com writes: >>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. > >So you're saying that having a Java runtime on the filer is an improvement >in security? That's insane. It would be insane; that's why I didn't say it. Your claim, which you neatly snipped, was: >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. To wit, my answer - it's not impossible. Whether or not other methods are more or less secure is a matter of debate, but your original point was wrong. >>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 hardwar >>e, >>although bad firmware could theoretically lead to physical damage of the disk >>drive hardware mechanism. > >It doesn't make it easier. It makes it trivial. This we'll just have to disagree on. We don't really know without statistically sampling over a number of security incidents and seeing exactly what the increased risk is, and even then we'd have to agree on the range of what is called "trivial". If you surveyed a group of informed security consultants on the issue, however, I don't think "trivial" would be the word of choice. >>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. > >And you will be unable to update it's /etc/passwd, /etc/quotas, etc. You >must not run a filer in an environment that changes often. Or when you do, you create a temporary security hole that you can monitor in realtime. (Any real security conscious admin are monitoring all of their stuff anyway, but that's another issue). The point is, again, your statement was wrong. Perhaps you were simply *overstating* to make a point... but in the world of security, one should be careful to convey accurate information, not hype. >>>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. > >It is true. Perhaps you should pull your head out of the sand for a >minute and stop blindly defending the existance a stupid command. By the way, I was only saying the first part of your statement (before the semicolon, that Netapp had raised the stakes) wasn't true; obviously from the context I agreed with the latter, but feel it's largely irrelevant since one should rate virutally the same possibility to previous OS version. You probably still disagree, but I wanted the clarification so we didn't go down a rathole over a misunderstanding. Even if you're correct, you're talking the filer's risk increasing only so much as to be in line with almost every other OS/platform out there that include facilities for such a thing. Hardly something to get worked up over. Bruce

1 0

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

12 Feb '99

>> still are some UNIX clients around whose NFS v3 implementations leave something >> to be desired on the stability and performance fronts. > >The question was about TCP, not the version of the NFS protocol. Sorry. That should have read "NFS v3 implementations and/or NFS vX on TCP implementations". >The reason for that choice had nothing to do with stability nor with >any client issues so far as I know, but rather because most customers >see a performance hit when running TCP with enabled, A performance hit which is oft blamed, sometimes deservedly sometimes not, on the quality of the client implementation. Arguably the mount(1M) command on the client side should favor the higher-performing-under-most-circumstances option (UDP), but NetApp doesn't control that so... Keith

2 1

← Newer
1
2
3
4
5
6
7
8
9
10
Older →

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 February 1999