toasters September 1998

toasters@lists.teaparty.net

55 participants
54 discussions

Re: Soft Links ON NT ( again)
by mds46523＠ggr.co.uk 23 Sep '98

23 Sep '98

On Sep 23, 10:59, Raymond Brennan wrote: > Subject: RE: Soft Links ON NT ( again) > Mark, > > I've got a quota tree on my F520 called "sandbox". My F520 system name is > "tap", mainly because I got fed up of typing in "netapp" all the time. > Anyway, what I'm trying to do is to get soft links with absolute path names > to work on NT. They work fine on Unix(as you woud expect). I've installed > SecureShare on all my PCs and am using Unix-style security on "sandbox" and > indeed as the default security style on all other trees. > > I use NIS on my LAN. "sandbox" on "tap" is accessed on Unix as > /eeg/sandbox-the NIS map is called auto.eeg and has this line in it : > > sandbox -intr tap:/sandbox > > So, on Unix, all I have to do is > > # cd /eeg/sandbox > > and I'm there. Hard links (ln file linkname) work fine on NT, as do relative > soft links ( ln -s ../file linkname). The problem I'm having is with soft > links with absolute path names on NT. If, for example, I do > # cd /eeg/sandbox/dir2 > # ln -s /eeg/sandbox/dir1/file linkname > > NT can't seem to access linkname on /eeg/sandbox/dir2 at all, apart from > showing it in the Windows Explorer. I hope this is clear. > > My /etc/symlink.translations file looks like this : > > Map /sandbox/* //sandbox/* > > The double slash on the second field is to counteract Bud ID 9553, which I > found on the NOW website. Also, I've tried terminating and restarting CIFS, > but without success. > > Thanks for any help you can give me. I think I know why the NT box is having problems, and I think I can demonstrate a similar situation on two Unix boxes. Simply put, absolute symlinks have too much information, they're too precise for either the NT box or the filer to map them correctly. To demonstrate my hypothetical scenario, I've use my home directory on ukwfs2 another F520 (running 4.3.1, for interest). On our (SunOS4.1.4) admin host: uk-dns-stv1-root /net/ukwfs2/users/d23/mds46523: ln -s \ /net/ukwfs2/users/d23/mds46523/dead.letter flibble uk-dns-stv1-root /net/ukwfs2/users/d23/mds46523: ln -s ./dead.letter flobble Creating two symlinks in my home directory to a simple text-file also in the directory, one absolute, one relative. Now, on a client, in my home directory: 0 ukwit12> head -2 flibble UX:head: ERROR: Cannot read from flibble 1 ukwit12> head -2 flobble >From mds46523 Thu Sep 17 18:07:40 1998 From: mds46523(a)ukwsv1.ggr.co.uk (Simmons Mr M D) The absolute symlink simply overspecifies the location, which my client has no way to map to something it can understand and use. The NT box is in as bad a position as my Unix client - the /eeg/... part of any absolute symlink in your Unix domain doesn't match anything in the NT box's wordlview, and the filer has no rule to map it for the NT box. Extend the mapping rules in the filer, or revert all absolute symlinks to relative symlinks and I think you'll be able to work around the problem. I'd try adding: Map /eeg/sandbox/* //sandbox/* to your /etc/symlink.translations file. > > :) > > "What do people mean when they say the computer went down on them?" > > --Marilyn Pittman > > >-----Original Message----- > >From: Mark Muhlestein [mailto:mmm@netapp.com] > >Sent: Tuesday, September 22, 1998 6:40 PM > >To: Raymond Brennan > >Subject: Re: Soft Links ON NT ( again) > > > > > >Send me a copy of your symlink.translations file, along with a specific > >description of what you are trying to do, and I'll try to help get you > >going. > > > >One thing that may not be documented clearly is that you have to restart > >cifs before the symlink.translations file is reloaded. Yes, it's a bug. > > > >Mark Muhlestein -- mmm(a)netapp.com > > > <SNIP> >-- End of excerpt from Raymond Brennan -- -Mark, TSG Unix admin and support, int 782 2499, ext +44 1438 76 2499. ... an Englishman in London ...

2 1

RE: Soft Links ON NT ( again)
by Raymond Brennan 23 Sep '98

23 Sep '98

Mark, I've got a quota tree on my F520 called "sandbox". My F520 system name is "tap", mainly because I got fed up of typing in "netapp" all the time. Anyway, what I'm trying to do is to get soft links with absolute path names to work on NT. They work fine on Unix(as you woud expect). I've installed SecureShare on all my PCs and am using Unix-style security on "sandbox" and indeed as the default security style on all other trees. I use NIS on my LAN. "sandbox" on "tap" is accessed on Unix as /eeg/sandbox-the NIS map is called auto.eeg and has this line in it : sandbox -intr tap:/sandbox So, on Unix, all I have to do is # cd /eeg/sandbox and I'm there. Hard links (ln file linkname) work fine on NT, as do relative soft links ( ln -s ../file linkname). The problem I'm having is with soft links with absolute path names on NT. If, for example, I do # cd /eeg/sandbox/dir2 # ln -s /eeg/sandbox/dir1/file linkname NT can't seem to access linkname on /eeg/sandbox/dir2 at all, apart from showing it in the Windows Explorer. I hope this is clear. My /etc/symlink.translations file looks like this : Map /sandbox/* //sandbox/* The double slash on the second field is to counteract Bud ID 9553, which I found on the NOW website. Also, I've tried terminating and restarting CIFS, but without success. Thanks for any help you can give me. :) "What do people mean when they say the computer went down on them?" --Marilyn Pittman >-----Original Message----- >From: Mark Muhlestein [mailto:mmm@netapp.com] >Sent: Tuesday, September 22, 1998 6:40 PM >To: Raymond Brennan >Subject: Re: Soft Links ON NT ( again) > > >Send me a copy of your symlink.translations file, along with a specific >description of what you are trying to do, and I'll try to help get you >going. > >One thing that may not be documented clearly is that you have to restart >cifs before the symlink.translations file is reloaded. Yes, it's a bug. > >Mark Muhlestein -- mmm(a)netapp.com > <SNIP>

1 0

Re: Soft Links on NT
by Keith Brown 22 Sep '98

22 Sep '98

>This mentions Release 4.3. However, I'm on 5.1D4 and does this work ? >In a word : NO. Any "bugs" of which I am unaware not withstanding... no! The CIFS symlink following mechanism should work in a Data ONTAP 5.1 release just as it did in the 4.3 releases. So let's see what might be up... >I've got an F520 running release 5.1D4. I've got a problem with reading Unix >soft links on NT. Symbolic links. OK... >NetApps Inc. say that soft links only work on NT within >one quota tree. That is a difficult thing t live with, but I suppose I'll >have to. The limitation that exists with the symlink following mechanism is that the encountered link must point to a location in the filer's file system that resides under the same *share* that the client encountering the link originally connected to in order to be able to hit the link in the first place (if you see what I'm saying? :-)). The "problem" is a rather fundamental one. If a link points to a location on the filer that is outside of the scope of the share being accessed from a client, our software has no way of telling the client to disconnect itself from the existing share and reconnect to another that could *possibly* allow it to go to the destination (even if one existed). Simply taking the client to the destination without regard to share security would be a "well iffy" piece of behavior from the security perspective. As for the quota tree limitation you specifically mention, yes, I believe that might be a new thing in 5.1, but there is a good reason for it and it isn't actually that much of a "new restriction" in practice. The good reason for it is that now our qtrees can have different security styles associated with them. You only get to inform a client about the security of the file system it is connecting to (FAT or NTFS) at connect time, so taking a client from a FAT file system (which actually has UNIX security on it in our case), to an NTFS file system would definately constitute a problematic situation in the eyes of the protocol. There's just no mechanism to cleanly handle such a transition. Oh... and the reason that it isn't actually that much of a (new) restriction is that qtrees are always rooted in the root directory of a volume, so the only share that a client could possibly connect to that would allow inter-qtree jumps while still obeying the "in the same share" rule, would be the C$, D$, E$ etc, etc... administrative shares, which aren't normally used by ordinary users. >I can make soft links with relative pathnames work OK, but what about links >with absolute pathnames ? On the NOW NetApp site, I found a sysadmin man >page mentioning the /etc/symlink.translations file, which is apparently a >way to make this work. It's a way to make *something* work, but not inter-share jumps. From a UNIX client you may have created symlinks that a filer could not possibly follow, because they point to locations that are not actually on the filer. A UNIX client can follow these links elsewhere because it is an NFS client. A filer can't, because it is not. The symlink translation mechanism lets you make a copy of the target data somewhere that *is* on the filer (manually), and then automatically translate the existing symlinks to hook a Windows client over to it. Make sense? Keith

1 0

Quad Fast Ethernet vs. Single Fast Ethernets
by Brian Tao 22 Sep '98

22 Sep '98

I've got two F740's on order, whose sole task will be to provide storage for a couple of Sun E450 Oracle boxes. I plan to have each filer attached to both E450's via crossover full-duplex 100baseT, plus a third 100baseT on the filer to the rest of the LAN. Is there any reason to go with the QFE board besides port density (which I don't really need?) Are there plans to do FastEtherChannel on it? The F740 has six available PCI slots, and I will likely not require more than the one on-board FC-AL controller (plus perhaps an additional one for clustering). Three single Fast Ethernet interfaces are cheaper than one QFE, failure of one only downs one interface, the load can be spread over more than one PCI bus (I don't know if this an issue on the Alpha boards) and they are cheaper to replace. OTOH, will the clustering software failover a filer if any one of its NIC's dies, and will that be support on both the single and quad NIC's? -- Brian Tao (BT300, taob(a)risc.org) "Though this be madness, yet there is method in't"

6 10

Re: Quad Fast Ethernet vs. Single Fast Ethernets
by Daniel Quinlan 22 Sep '98

22 Sep '98

Val Bercovici <valb(a)netapp.com> writes: > Unfortunately, one of the things we don't automatically trigger a > failover on in our first Clustered Failover release will be a "normal" > interface link failure (OTOH, if a NIC fries its PCI slot, we will > failover). Using a Cisco Fast EtherChannel across at least two > interfaces would be a nice way to protect yourself from a single > interface failure on a filer. We were hit by an interface link failures once last year. What options are there or will there be (either with or without Clustered Failover) to offer failover or redundancy for failed interfaces? I'm especially concerned about this as we plan to transition from one or two quad fast ethernet cards on each filer to one or two gigabit ethernet cards per filer. Thanks. - Dan

6 10

Soft Links ON NT ( again)
by Raymond Brennan 22 Sep '98

22 Sep '98

Hello, Look what I found in the mailist list archive : From madhatta Tue Nov 18 18:52 EST 1997 Received: from odin.netapp.com (odin.netapp.com [198.95.224.183]) by turing.mathworks.com (8.8.5/8.7.3) with ESMTP id SAA03647 for <toasters(a)mathworks.com>; Tue, 18 Nov 1997 18:51:35 -0500 (EST) Received: from herra.netapp.com (172.20.6hide.netapp.com [198.95.224.200]) by odin.netapp.com (8.8.7/8.8.7/GNAC-GW-2.1) with ESMTP id PAA11514; Tue, 18 Nov 1997 15:44:32 -0800 (PST) Received: from netapp.com (keith-pc.netapp.com [172.20.6.61]) by herra.netapp.com (8.8.7/8.8.7/GNAC-GW-2.1) with ESMTP id PAA26897; Tue, 18 Nov 1997 15:47:38 -0800 (PST) Message-ID: <3472290E.C00B8959(a)netapp.com> Date: Tue, 18 Nov 1997 15:47:27 -0800 From: Keith Brown <keith(a)netapp.com> Reply-To: keith(a)netapp.com Organization: Network Appliance, Inc. X-Mailer: Mozilla 4.02 [en] (Win95; I) MIME-Version: 1.0 To: quinlan(a)transmeta.com CC: toasters(a)mathworks.com Subject: Re: links using CIFS References: <199711182300.PAA22605(a)sodium.transmeta.com> Content-Type: multipart/mixed; boundary="------------FBDEBC85A108A84A65F02E8D" Content-Length: 2731 X-Keywords: Status: RO X-Status: This is a multi-part message in MIME format. --------------FBDEBC85A108A84A65F02E8D Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Daniel Quinlan wrote: > Could someone explain exactly what works and what doesn't when using > symbolic links from a CIFS client (Windows95 or WindowsNT, primarily) > from a Netapp with Unix symbolic links? Certainly. In Data ONTAP releases prior to 4.3, Windows systems saw symbolic links that had been created in the WAFL file system (from the UNIX side) as exactly what they *really* are. That is to say, plain files which contain the full or relative pathname to the place that they are a symbolic link to. Technically, the UNIX symlink attribute has no meaning in the Windows client world, and so they don't treat symlink files as "special" in the same way that UNIX systems do. However, in our brand spanking new 4.3 release, this has changed. If you install and run this release on your filer(s), you will find that Windows systems will now automagically follow UNIX symlinks, with a single, necessary limitation. The symlink must point to a place that the client *can* actually make the jump to. The client can only be redirected to another location under the same share in which it encountered the symlink in the first place. Just in case you have designed your symlinks in such a way that this is not entirely feasible for you to easily arrange, the new mechanism also features a symlink pathname translation mechanism that works very simililarly to the existing "httpd.translations" mechanism with which customers may already be familiar. You can use this mechanism to yank meandering symlinks back into your share, but of course you probably have to put a copy of the data that they were originally going to take you to in the share also. Ooh.. I should probably mention that the new symlink following feature does contain machinery to stop a Windows client from looping in a file system when it is following the symlinks. That base is covered! :-) Keith This mentions Release 4.3. However, I'm on 5.1D4 and does this work ? In a word : NO. Yours beleaguredly, Raymond the Depressed "Your future depends on your dreams. So go to sleep "- Lao Tse

1 0

Soft Links on NT
by Raymond Brennan 22 Sep '98

22 Sep '98

Hello All, I've got an F520 running release 5.1D4. I've got a problem with reading Unix soft links on NT. NetApps Inc. say that soft links only work on NT within one quota tree. That is a difficult thing t live with, but I suppose I'll have to. I can make soft links with relative pathnames work OK, but what about links with absolute pathnames ? On the NOW NetApp site, I found a sysadmin man page mentioning the /etc/symlink.translations file, which is apparently a way to make this work. Well, I've tried it and for the life of me I can't get it to work. Anyone else have a similar experience with this ? Raymond Brennan mailto: ray_brennan(a)mentorg.com "Children in back-seats cause accidents; Accidents in back-seats cause children!" -Ike Turner

1 0

Re: Quad Fast Ethernet vs. Single Fast Ethernets
by Daniel Quinlan 22 Sep '98

22 Sep '98

kls(a)netapp.com (Karl Swartz) writes: > On the other hand, there probably are a lot of customers who would > rather have failover *now* even if it only works with 100base-TX or > FDDI (or requires extra GbE NICs). If we told a FDDI customer that > we'd be able to give them failover today but we won't because we've > not finished a driver for a GbE NIC that can handle multiple MAC > addresses, they'd probably be a bit peeved. I agree about "failover *now*". FDDI -- isn't it dead yet? ;-) > So, your options at the moment are > (a) use GbE but not have failover > (b) use failover but not GbE > (c) buy a second GbE NIC and have both failover and and GbE > (d) wait until we have a GbE NIC and driver which will allow you > to have failover with a single GbE What about GbE and failover to non-GbE (fast ethernet)? Just to make sure, does failover require EtherChannel? - Dan

2 1

Log rotation.
by just me. 22 Sep '98

22 Sep '98

Anyone know of a knob i can turn to adjust the log rotation interval, or, alternately, a way to turn it off? I'd much rather have my own script doing this, but don't want to mess around with race conditions. Thanks. matto --matt(a)snark.net----------------------------------------------------- Matt Ghali MG406/GM023JP - System Administrator, interQ, Inc AS7506 "Sub-optimal is a state of mind." -Dave Rand, <dlr(a)bungi.com>

1 0

subscribe : toasters
by Ron Neuenberger 21 Sep '98

21 Sep '98

1 0

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 September 1998