toasters

toasters@lists.teaparty.net

13515 discussions

RE: syslog
by Hackworth, Brian 17 Mar '99

17 Mar '99

: Is there any way to get syslog to log not only a telnet : connection (default) but also rsh connections and the : command that was executed? The ONTAP rshd does not log successful logins, nor does it log the command that was executed. We're considering this feature for a future release. : Out of the box, there isn't an etc/syslog.conf file, : only a sample one. It appears to be logging: If there's no syslog.conf file, ONTAP uses the equivalent of *.info /dev/console *.info /etc/messages : How do I "kick" ONTAP to start using a new syslog.conf file? : "Reboot the filer" is not an acceptable answer. ;) No action is necessary. ONTAP notices the change and incorporates it automatically. - Brian Hackworth

1 0

CIFS directory conversion of NFS created files
by Todd C. Merrill 17 Mar '99

17 Mar '99

This from the docs for ontap 5.2.1: file://....5_2_1-cd/ontap/5_2_1/doc/html/sag/multi5.htm#1180418 Timing of directory conversion for CIFS access to NFS-created files [...] One-time access conversion If you add CIFS access to files that were created by NFS clients, the first time a directory is accessed through CIFS, the filer automatically converts it to the format required to support CIFS clients. "Accessed" here includes listing the contents of a directory; for example, with the File Manager, Explorer, or DOS dir command, or by accessing any files or directories contained in that directory. Under some circumstances, the CIFS client might automatically look at the directory's subdirectories, thereby triggering conversion of the subdirectories as well. (For example, this happens if the user has set the Expand All option under the File Manager Tree menu.) After a directory is converted, it does not have to be converted again. Here's my question(s): I have a directory on the filer root (at the level of etc) called "devel." Under that is a tree called "tree." A CIFS share is defined on "tree." Under "tree" is a bunch of other directories: devel/tree/{dir1,dir2,dir3,...dirN} When I first got the filer, I did a right-click:properties from an NT box on all directories dir1,dir2,...dirN in order to force the conversion process recursively. Things have been running smoothly. Until yesterday. The darn "access is denied" error message on the NT box is showing up again under the nightly snapshot for this tree on some directories. I suspect the directories are new, but have been renamed to previously named directories. (I.e., `mkdir new; mv dir old; mv new dir; rm -rf old` might have happened from an NFS mount). Question1: If I create a new directory devel/tree/dirnew and put some stuff in it, and no NT box access it before the nightly snapshot, this directory will give an "access denied" message to any NT box trying to access this directory from the snapshot, right? (I believe the answer is "yes" since this seems to jive with empirical data.) Question2: If I define a share at "devel" and then convert the entire *tree* directory, instead of dir1,dir2,...dirN, will any subsequent directory created under that tree (at the level devel/tree/dirnew) already automatically be converted, since its parent was? (This is kinda like the old SunOS subdirectories inheriting the group (?) owner of the parent.) (I think the answer is no, from FAQ#486). Question3: If the answer to #2 is no, is there another way to make this happen automagically? It seems reasonable, especially if a share is defined on any directory higher in the tree. Until next time... Todd C. Merrill The Mathworks, Inc. 508-647-7792 24 Prime Park Way, Natick, MA 01760-1500 508-647-7012 FAX tmerrill(a)mathworks.com http://www.mathworks.com ---

1 0

RE: filer NFS to hp700 or alpha: client cache problem?
by Thomson, Michael (UK, GIS ISM Group for R&D) 17 Mar '99

17 Mar '99

Todd, Could I suggest you check the time synchronisation between your client box (doing the linking) and the Filer? In a previous life I found that 'make' and other tools ignore files if their creation dates are in the 'future', even if it's only a few seconds to a minute. I suspect that might explain why depending on how quickly you run the 'rm', only 2/3 or all the files vanish. Now, all together: "Please can we have Network Time Protocol support on our filer?" Cheers, Mike > -----Original Message----- > From: Todd C. Merrill [SMTP:tmerrill@mathworks.com] > Sent: Tuesday, March 16, 1999 10:43 PM > To: toasters(a)mathworks.com > Subject: filer NFS to hp700 or alpha: client cache problem? > > > We haven't narrowed down the problem entirely yet, but I'd thought I'd > ask anyway.... > > During a linking stage, in linking a bunch of .o files into a library > .a, the linker will grab about 2/3 of the files in a directory into the > archive, and that's it. A subsequent `rm` of *.o sometimes removes only > those exact 2/3 files, sometimes removes all of them, even though the > archive only has 2/3 of them in it. > > It seems as if the directory inode is only showing the oldest 2/3 of the > files, and depending on how long it takes the `ar` and `rm` commands to > execute, removes 2/3 or all of the files. > > I don't suspect it's the filer; watching an `ls` of the directory from > another machine (Solaris) shows all the files there. There are only > about 150 .o files to archive. > > Anyone aware of any client-side NFS caching problems with Alpha or HP700 > boxes? The HP is running HPUX10.20 and appears to be using only NFS V2; > the alpha is running Digital UNIX V4.0D and appears to be using only V3. > (NFS version from `nfsstat -h hostname` on the filer.) The filer is set > nfs.tcp.enable off, NetApp Release 5.2.1. > > Until next time... > > Todd C. Merrill > The Mathworks, Inc. 508-647-7792 > 24 Prime Park Way, Natick, MA 01760-1500 508-647-7012 FAX > tmerrill(a)mathworks.com > http://www.mathworks.com > ---

2 1

Backups from DUMP command and snapshots
by Mike Ball 17 Mar '99

17 Mar '99

Questions Please understand I am a novice to the "dump" command when answering the following questions. Thank you for the help! 1) From what I understand running, a full backup with the "dump" command does not backup existing snapshots. Is this correct? 2) It looks like if you want to backup snapshots, you must backup a particular snapshot one at a time. Is this correct? 3) If I do a full restore (if the file system on the filer completely crashed) from a backup created by the "dump" command on the filer, it looks like I must first create the exact qtree structure which was previously on the filer before I can run the full restore. Is this correct?

2 1

filer NFS to hp700 or alpha: client cache problem?
by Todd C. Merrill 17 Mar '99

17 Mar '99

We haven't narrowed down the problem entirely yet, but I'd thought I'd ask anyway.... During a linking stage, in linking a bunch of .o files into a library .a, the linker will grab about 2/3 of the files in a directory into the archive, and that's it. A subsequent `rm` of *.o sometimes removes only those exact 2/3 files, sometimes removes all of them, even though the archive only has 2/3 of them in it. It seems as if the directory inode is only showing the oldest 2/3 of the files, and depending on how long it takes the `ar` and `rm` commands to execute, removes 2/3 or all of the files. I don't suspect it's the filer; watching an `ls` of the directory from another machine (Solaris) shows all the files there. There are only about 150 .o files to archive. Anyone aware of any client-side NFS caching problems with Alpha or HP700 boxes? The HP is running HPUX10.20 and appears to be using only NFS V2; the alpha is running Digital UNIX V4.0D and appears to be using only V3. (NFS version from `nfsstat -h hostname` on the filer.) The filer is set nfs.tcp.enable off, NetApp Release 5.2.1. Until next time... Todd C. Merrill The Mathworks, Inc. 508-647-7792 24 Prime Park Way, Natick, MA 01760-1500 508-647-7012 FAX tmerrill(a)mathworks.com http://www.mathworks.com ---

4 3

RE: Etherchannel (was Re: NetApp/Auspex killer?)
by Mohler, Jeff 16 Mar '99

16 Mar '99

FYI here is what I have on EC on the NA. *heh* <<TRUNKING.html>> > -----Original Message----- > From: Bennett Todd [SMTP:bet@newritz.mordor.net] > Sent: Tuesday, March 16, 1999 11:48 AM > To: tkaczma(a)gryf.net > Cc: toasters(a)mathworks.com > Subject: Etherchannel (was Re: NetApp/Auspex killer?) > > 1999-03-16-02:01:42 Tom: > > Speaking of EtherChannel or more generically Link Aggregation. Does > > anyone know a link describing it in quite a detail without going through > a > > complete protocol suite definition. > > Don't know of a link, would love to see one. But as I understand it, the > general strategy is that very roughly each side separately runs simple > algorithms to attempt to schmeer traffic over the multiple links. Often > the > algorithm is based on the MAC addrs of the two endpoints, and so won't > yield > over 100Mbps for any single link, but will hopefully (3 times out of 4) > place > competing high-volume connections on separate links. > > There are then hacks for failing over to another line when one link drops, > and > other border-condition exotica. > > But the basic idea is that given a large number of communication > endpoints, > you can schmear the traffic over multiple links quite cheaply. > > Dunno if this is any help to you. > > -Bennett

1 0

Re: BDC/NAC issues
by Mark Muhlestein 16 Mar '99

16 Mar '99

Not true. If a BDC the filer is talking to goes away the filer will just attempt to connect with another one. If the "NAC people" are NetApp employees, please let us know who they are so we can give them the correct information. Mark Muhlestein -- mmm(a)netapp.com -----Original Message----- From: Graham C. Knight <grahamk(a)ast.lmco.com> To: toasters(a)mathworks.com <toasters(a)mathworks.com> Date: Tuesday, March 16, 1999 10:13 AM Subject: BDC/NAC issues >Someone just posted this message to just about >every top level manager in our company :-( > >> >> What I am hearing 2nd hand is that NAC people say that the NAC's get locked >> into a BDC for authentication. >> If that BDC goes away or gets renamed that all the NACs will need to be >> removed from server manager, CIFS will have to be reconfigured, and then >> they will have to be added back into the servers Domain. >> >> If this is true then we have no unattended failover capabilities for a BDC >> dying. This could be a significant problem. > > >So - is this true? I don't know enough about NT to comment... > >TIA, > Graham

1 0

BDC/NAC issues
by Graham C. Knight 16 Mar '99

16 Mar '99

Someone just posted this message to just about every top level manager in our company :-( > > What I am hearing 2nd hand is that NAC people say that the NAC's get locked > into a BDC for authentication. > If that BDC goes away or gets renamed that all the NACs will need to be > removed from server manager, CIFS will have to be reconfigured, and then > they will have to be added back into the servers Domain. > > If this is true then we have no unattended failover capabilities for a BDC > dying. This could be a significant problem. So - is this true? I don't know enough about NT to comment... TIA, Graham

1 0

Any comments on emergin SANS technology...
by Christoph Doerbeck A242369 16 Mar '99

16 Mar '99

Just wondering if anyone on the list has an opinion on Storage Array Networks (SANS)... Should NetApp try to develop another product line for SANS based on WAFL? I don't know enough about the underlying technology to make any clever or intelligent comments. - Christoph

1 0

5.2.1D5: praise and a caveat
by Brian Rice 16 Mar '99

16 Mar '99

Last Friday night I put 5.2.1D5 (the successor to 5.2.1P1) on one of the F630's. It didn't panic, and I've got my quotas back. Ahhhh, yeah. Just one leetle caveat, one which caused me to believe I had a hung filer and to place a down-filer call to NetApp technical support. It seems that 5.2.1D5 sets the floppy-boot? environment variable for you by default. So when you boot, you see something like... NetApp Release 5.2.1D5: Mon Mar 8 23:54:01 PST 1999 Copyright (c) 1992-1998 Network Appliance, Inc. Starting boot on Sat Mar 13 07:50:19 GMT 1999 Scanning for disk drives: floppy boot? .......................... Spinning up disk drives: 8a.12 8b.12 8a.13 8b.13 8a.14 8b.14 Configuring disk drives: 8b.14 8a.14 8a.9 8a.12 8b.11 8b.12 8b.9 8b.8 8b.4 8b.5 8b.1 8a.3 8a.10 8a.2 8a.11 8a.5 8b.0 8b.10 8b.13 8b.2 8b.3 8a.13 8a.8 8a.1 8a.0 8a.4 and then, apparently, the filer hangs, with the LCD saying "Configuring disk 8a.4" (or whichever one was last). Turns out that that "floppy boot?" way up there at the top was a prompt. It's waiting for you to type "yes" or "no" on the console. A "no" causes the boot to finish normally. Then a "setenv floppy-boot? false" and a reboot gives you a happy filer. Thanks to the NetApp tech support guy for reacting so graciously to being beeped in the middle of his Friday night. Brian

3 3

Re: NetApp/Auspex killer?
by Darrell Fuhriman 16 Mar '99

16 Mar '99

discussing hardware completely ignores the core of what makes the Netapp so good -- the file system. WAFL simply beats all the others -- even ext2 or veritas. Darrell

1 0

Re: 5.2.1D5: praise and a caveat
by Aimee Schmidt 15 Mar '99

15 Mar '99

> That's strange. My F760 doesn't have a "floppy-boot?" parameter. What it > does have is a "boot" (or is it "system-boot"?) parameter set to > "floppy scsi-disk". That way my box boots from floppy, if installed, otherwise > from disk. Is this different in 5.2.1D5? The boot order should be controlled > by the firmware (2.1_a2) not ONTAP. (after all ONTAP isn't running yet) > > I cannot check my 630's (firmware 1.12_a1) but I suspect they are the same. This variable is not set by default. One must execute a setenv floppy-boot? true at the firmware level (ok prompt). Conversly one can remove the prompt with the following syntax at firmware: unsetenv floppy-boot? Aimee _____________________________________________________ Aimee Schmidt aimee(a)netapp.com Software Test Engineer Voice: 408.367.3238 Network Appliance Fax: 408.367.3451 / |\ / | \ "We cannot direct the wind / | \ but we can adjust our sails" ---------- \________/ |_|

1 0

Re: 5.2.1D5: praise and a caveat
by grdodson＠lexmark.com 15 Mar '99

15 Mar '99

That's strange. My F760 doesn't have a "floppy-boot?" parameter. What it does have is a "boot" (or is it "system-boot"?) parameter set to "floppy scsi-disk". That way my box boots from floppy, if installed, otherwise from disk. Is this different in 5.2.1D5? The boot order should be controlled by the firmware (2.1_a2) not ONTAP. (after all ONTAP isn't running yet) I cannot check my 630's (firmware 1.12_a1) but I suspect they are the same. Thanks. Graydon Dodson grdodson(a)lexmark.com Lexmark International Inc.

1 0

RE: filer-mrtg
by Thompson, Tim 15 Mar '99

15 Mar '99

> > > In a fit of needing a diversion i installed > > > filer-mrtg yesterday and boy is it cool! > > Glad you found it. > > Ok, I looked on ftp.netapp.com, But I can't find it. > > Where can you get filer_mrtg? On the NOW site (now.netapp.com) in the Tools&Utilities section, in the Toolshed area. (You could also use the "Search" capability on the NOW site.) >From the README: This package uses the MRTG (Multi Router Traffic Grapher) to provide web-based monitoring for Filers. It provide a current and historical view (the last day, week, month, and year) of statistics for network traffic (per interface), CPU load, disk usage, and NFS and CIFS operations. There is also a similar netcache-mrtg package in the Toolshed. ...Tim Thompson...Network Appliance, Inc...tjt(a)netapp.com...

1 0

Filers and Lotus Notes...
by Paul Taylor 15 Mar '99

15 Mar '99

Hi. I would like to know if anyone has had any experience with NetApp filers and Lotus Notes. Here is our situation: We have an F630 that has 2 LN server attached to it (among other things), running across Fast Etherchannel. One of the LN servers is on WinNT and the other is on AIX. The AIX box is a RS/6000 H50 with a single CPU, and 512MB RAM. The NT box is a single CPU netfinity with 512MB RAM, running at 400mhz. The Filer has 512MB onboard, running DoT 5.1D12. LN is v. 4.6.2. We have 2 networks on each server, a public network that goes to the internet, and a private network (a 10. ) that goes to the filer. The problem we had at first was with LN itself not liking UNC names. For this problem I installed PCNFSD, and ran it that way for a while. Now we are experiencing corruption at random times. I figure it is probably the connectionless nature of NFS. Bah. Anyone have any better ideas? BTW: client doesn't want to do directory links, it has to run as a service, and it has to run as system. Next problem: We installed Domino on an RS6K adn it ran horribly. I moved it over to local disk, and it ran great. Hmmmmm.... Here is the mount info for the rs6k from /etc/filesystems (commented out for now since we went to local disk: * /lnmobem01: * dev = "/vol/vol0/lnmobem01" * vfs = nfs * nodename = efiler01 * mount = true * options =bg,hard,intr,rsize=4096,wsize=4096,vers=2,proto=udp account = true Anyone have any ideas as to why this would act slow when opening a DB? it takes about 1.5 min to open log.nsf, when it should take about .5 sec. Thanks, Paul

1 0

RE: filer-mrtg
by Thompson, Tim 15 Mar '99

15 Mar '99

> In a fit of needing a diversion i installed > filer-mrtg yesterday and boy is it cool! Glad you found it. >I'm having a hard time >correlating the "Network Kbytes sent >and recieved" with the "Network Interface >Sent And Received" The first one is uses the Filer's MIB entry for "miscNetRcvdKB", and the second one uses the MIB entry for the individual interface. If only one interface is used, they should match - I just checked a filer here that has a single interface, and the two graphs match well. I'll ask for more details in separate email. ...Tim Thompson...Network Appliance, Inc...tjt(a)netapp.com...

2 1

Etherchannel/trunking
by Paul English 15 Mar '99

15 Mar '99

We currently have a quad etherchannel/trunking set up working with a Cisco switch, and from Todd's post, you all know that it worked very well. Does anyone have any experience using etherchannel/trunking with any other vendors/products? Any info would be useful.. Paul -- Paul English (Tallpaul) (508) 647-7715 tallpaul(a)mathworks.com (508) 647-7015 FAX Walkie Channel 2

4 3

NIS+
by Jason Middlebrooks 15 Mar '99

15 Mar '99

I know that the netapps support NIS but what about NIS+ ????? Thanks for all the answers. Jason

3 2

filer-mrtg
by Graham C. Knight 12 Mar '99

12 Mar '99

In a fit of needing a diversion i installed filer-mrtg yesterday and boy is it cool! I do have a question for those in the know though - i'm having a hard time correlating the "Network Kbytes sent and recieved" with the "Network Interface Sent And Received" I would figure that on a system with 1 interface that these graphs should be pretty much identical - but on some systems the path of the graph is very similar but the numbers are really different - sometimes as much as 2-1 in favor of the "Network Kbytes Sent and Recieved" category. Is this a bug - or am i missing something here.... Thanks for any info you all can provide. GRaham

2 1

RE: FW: Shelves, disks, et al
by Chen, Ray 12 Mar '99

12 Mar '99

Probably because we qualified both. Dual-sourcing critical components like disk drives is standard procedure when multiple sources are available. (I can't say for sure because the qual for those happened before I got here.) Doing so has saved the posterior anatomy of more than one system vendor when it turns out that the product delivered by one source suddenly develops problems due to manufacturing process changes that occurred after qualification and/or problems that take a long time to appear (anyone remember the IBM stiction problems :-), or availability simply becomes a problem. I would imagine that most vendors don't want to qualify more than 2 both due to the cost of each qual and also because of the fact that qualifying N drives turns into an N^2 problem. If you qualify 3 drives, you have to ensure that all combinations work (e.g. a SCSI bus with drive A, B, and C works well, ditto for just drive A, B, and drive A, C, etc.). You wouldn't think that this should be an issue but it has been in the past. I think you can get away with skimping on some testing for situations where the N Gbytes and M Gbyte drives from the same manufacturer are the same family running the same firmware. But even then, I suspect you have to check for vibration issues in your supported enclosures. Ray > -----Original Message----- > From: Jonah Yokubaitis [mailto:barron@staff.texas.net] > Sent: Thursday, March 11, 1999 7:18 PM > To: Chen, Ray > Cc: 'Jim Davis'; toasters(a)mathworks.com > Subject: RE: FW: Shelves, disks, et al > > > > On Thu, 11 Mar 1999, Chen, Ray wrote: > > |The truly sad thing is that you are correct in that > |most people won't buy Michael's argument but that > |Michael is correct when he tries to communicate > |how much work goes into disk drive qualification > |and how painful the problems can be if you try > |and put in joe-random-disk drive, even the same > |model from the same vendor, and get unlucky. > | > |Heck, *I* didn't buy the argument until I spent > |a few years at another company where I exposed to > |some of what happens in drive qualification and > |saw what could go wrong in the lab and in the field. > | > |To be safe, you really need the same vendor, same > |drive, and exact same firmware and that's typically > |very hard to get unless you buy from the system vendor > |instead of the drive vendor. Even then, that's not > |a guarantee that you won't get bit, your odds are > |just better. > > Ray, > > If that's the case, then why does one of our F630s (9gig SCSI drives > in it) having 2 *different* model numbers of drives in it? All drives > came directly from Netapp. > > Jonah > > Jonah Barron Yokubaitis > http://www.texas.net <--- The ISP > http://www.giganews.com <--- News Outsourcing >

1 0

Re: CIFS share access?
by Allan Burrington 12 Mar '99

12 Mar '99

Assuming the filer can handle standard PC functionality - you can use subst to map any path to a drive letter. Target has share: C$ Within the C$ share is a directory named 'foo'. Do: subst z: \\target\c$\foo You now have z: rooted at the foo directory. -A ----- Original Message ----- From: Keith Brown <keith(a)netapp.com> To: Alexander Strauss <strauss(a)ti.com>; <toasters(a)mathworks.com> Sent: Thursday, March 11, 1999 10:27 AM Subject: Re: CIFS share access? >>I shared the directory where most of our Unix-Homes are located under >>the CIFS sharename UNIX_HOMES. There's no problem to connect >>\\filer\unix_homes, but I cannot connect directly to a sub-directory, >>i.e. \\filer\unix_homes\alex. > >As has already been stated, the cifs.home_dir option of Data ONTAP might >provide you with the functionality that you are looking for. > >However... I suspect that you *might* be under the impression that CIFS shares >are supposed to work like NFS exports in the above respect. Alas they don't. > >As you probably know, if you export "/home/keith" via NFS, it makes it possible >to directly mount any subdirectory of "/home/keith" from an NFS client (e.g. >"/home/keith/docs/cifs" or somesuch). Shares aren't like that. If you share >"/home/keith" as "KEITHDIR", all that you can connect to from a CIFS client is >"\\FILERNAME\KEITHDIR". You can't do something like a: > > net use q: \\FILERNAME\KEITHDIR\docs\cifs > >Keith >

2 1

Re: CIFS share access?
by Mark Muhlestein 11 Mar '99

11 Mar '99

You are correct, Alan; this works fine. Thanks for the good pointer. Mark Muhlestein -- mmm(a)netapp.com -----Original Message----- From: Allan Burrington <allan(a)mirapoint.com> To: Keith Brown <keith(a)netapp.com>; Alexander Strauss <strauss(a)ti.com>; toasters(a)mathworks.com <toasters(a)mathworks.com> Date: Thursday, March 11, 1999 11:48 AM Subject: Re: CIFS share access? >Assuming the filer can handle standard PC functionality - you can use subst >to map any path to a drive letter. > > Target has share: C$ > Within the C$ share is a directory named 'foo'. > >Do: > subst z: \\target\c$\foo > >You now have z: rooted at the foo directory. > >-A > >----- Original Message ----- >From: Keith Brown <keith(a)netapp.com> >To: Alexander Strauss <strauss(a)ti.com>; <toasters(a)mathworks.com> >Sent: Thursday, March 11, 1999 10:27 AM >Subject: Re: CIFS share access? > > >>>I shared the directory where most of our Unix-Homes are located under >>>the CIFS sharename UNIX_HOMES. There's no problem to connect >>>\\filer\unix_homes, but I cannot connect directly to a sub-directory, >>>i.e. \\filer\unix_homes\alex. >> >>As has already been stated, the cifs.home_dir option of Data ONTAP might >>provide you with the functionality that you are looking for. >> >>However... I suspect that you *might* be under the impression that CIFS >shares >>are supposed to work like NFS exports in the above respect. Alas they >don't. >> >>As you probably know, if you export "/home/keith" via NFS, it makes it >possible >>to directly mount any subdirectory of "/home/keith" from an NFS client >(e.g. >>"/home/keith/docs/cifs" or somesuch). Shares aren't like that. If you share >>"/home/keith" as "KEITHDIR", all that you can connect to from a CIFS client >is >>"\\FILERNAME\KEITHDIR". You can't do something like a: >> >> net use q: \\FILERNAME\KEITHDIR\docs\cifs >> >>Keith >> > >

1 0

Re: NFS fundamentally broken ??? (HA implications)
by mds＠gbnet.net 11 Mar '99

11 Mar '99

On Wed 10 Mar, 1999, Philip Thomas <thomas(a)act.sps.mot.com> wrote: > Hi, > An experienced sys admin and a colleague of mine argue that > NFS is "fundamentally broken", in the sense that because of > it's stateless architecture, a NFS client can "hang" during > shutdown if it's NFS server(s) does not respond. > This is aggravated when an "ack" is lost between server/client. > He claims to have gone through NFS code while he was at a prestigious university. > Granted I am not an expert, all the years I have worked with NFS, > I never had a problem shutting down a system because it's NFS > mount was not "responding". > But I am hoping for folks with "theoretical" background to shed some > more light into the above argument. > Both from experience, and from what I know of the workings of NFS, it *is* possible for a client to hang for valid reasons, even when shutting down, on mounts from unresponsive NFS servers. You can work around this by using dangerous options like soft mounts, or by killing processes that are using the NFS mounted areas and forcing the client to umount problem mounts. Excitingly, NetApp's use clever status-monitoring such that they can maintain NFS locks (even though they're just advisory) when the server reboots, and they, as servers, can suffer from clients that have died away with open locks. NetApp's spend some time every few minutes looking for such clients, and slowing down a bit while they do so. They also syslog a lot about that sort of thing. So there's a bit of a two-way street here, even though NFS is putatively stateless. We recently hit a bug related to that and a very helpful NetApp engineer (who I won't name to save him any embarrassment) is looking at how to smooth out that sort of wrinkle in the future - in fact at least part of that is going into a patch release in the near future. Check out the differences between intr and non-intr, and soft and hard mount options in the man-pages, and Hal Stern's book - I think that might make things clearer. > This brings to the second question. > The same person emphasize the above issue as an added reason for not > having NFS mount points on his HP High Availability (ServiceGuard). > Is there any site out there, in this wide world, using HP's ServiceGuard > (or any other HA) on a system that is an NFS client and would like to > comment on their experience in private or in this forum. Ah. There he has a point in principle, but in practicality I'd expect there to be no particular issue. We run a couple of Sun HA1.3 clusters here, and there's a Sun Cluster 2.x going in that I've had some sight of. They all use NFS mounts to import filesystems from servers, both Unix and NetApp. The design of HA clusters basically hinges on timeouts (as NFS does itself, thinking about it). If a node in an HA cluster must go down, it *will*, even if it's hanging on NFS mounts from elsewhere - usually because one timeout or other will trigger it's suicide, or for a sibling to commit fratricide (sororicide?). Obviously suicide or fratricide leaves a machine open to problems where neat and tidy shutdown without hanging on NFS mounts will be far superior. The fact that one node is having problems with an NFS mount probably means the other node(s) will too, and that will impact the overall HA cluster, which is bad(tm). So the secret is to make sure any mounts in an HA cluster are (a) required absolutely for the function the cluster provides and (b) are at least as reliable as the cluster is intended to be. I'm happy to use our NetApp's fs's on our HA clusters. I happen to think the group implementing the Cluster 2.x environment are storing up problems for themselves by using Unix NFS servers. Your mileage will vary. Hope that's helpful. > > Philip Thomas > Motorola - PEL, M/S M350 > 2200 W. Broadway M350 > Mesa, AZ 85202 > rxjs80(a)email.sps.mot.com > (602) 655-3678 > (602) 655-2285 (fax) >-- End of excerpt from Philip Thomas

2 1

Re: CIFS share access?
by Keith Brown 11 Mar '99

11 Mar '99

>I shared the directory where most of our Unix-Homes are located under >the CIFS sharename UNIX_HOMES. There's no problem to connect >\\filer\unix_homes, but I cannot connect directly to a sub-directory, >i.e. \\filer\unix_homes\alex. As has already been stated, the cifs.home_dir option of Data ONTAP might provide you with the functionality that you are looking for. However... I suspect that you *might* be under the impression that CIFS shares are supposed to work like NFS exports in the above respect. Alas they don't. As you probably know, if you export "/home/keith" via NFS, it makes it possible to directly mount any subdirectory of "/home/keith" from an NFS client (e.g. "/home/keith/docs/cifs" or somesuch). Shares aren't like that. If you share "/home/keith" as "KEITHDIR", all that you can connect to from a CIFS client is "\\FILERNAME\KEITHDIR". You can't do something like a: net use q: \\FILERNAME\KEITHDIR\docs\cifs Keith

1 0

FW: Shelves, disks, et al
by Alvarado, Michael 11 Mar '99

11 Mar '99

-----Original Message----- From: Stephen C. Woods [mailto:scw@sprynet.com] Sent: Wednesday, March 10, 1999 12:45 PM To: toasters(a)mathworks.com Subject: Shelves, disks, et al I've got a couple of questions for the list. (1) Can 9 GB drives be used in the same type shelves as 4 GB drives? I'm talking about narrow drives here. (2) Can they be mixed (4 and 9 GB drives) in a shelf? How about on the same controller? (3) Is there any source for drives other than Netapps? (4) If one obtains drives from another source can they be put under support (yeah I know, fat chance)? Would they void current hardware support? Would they void software support? (5) If I had some extra narrow shelves is there a market? We're thinking about decommissioning one of our older filers. Reply-to set to the list, as I'll bet dollars to doughnuts that other folks are interested. Thanks in advance Steve Woods ******************************************************************** Hi Steve and to the other members of the toasters(a)mathworks.com list I am happy to answer the questions you posed about Network Appliance storage Regarding the question about using non-Network Appliance disk drives in our filers. Our policy is that only Network Appliance-supplied disk drives can be used in our filers else the warranty is voided on the filers. There are very good reasons behind this policy. 1. Extensive screening and thorough integration is the difference. Network Appliance-supplied drives are burned in, carefully integrated, and screened to minimize reliability problems and maximize uptime. 2. Special driver software. Network Appliance drives have special driver software that has been written to optimize the performance of drives and filers. Every disk drive has unique characteristics. We carefully evaluate those characteristics and we specify specific changes to ensure the best possible performance. 3. Must meet exacting specifications. Drives are often seen as indistinguishable and interchangeable. This is not the case. 9 GB drives can differ by power consumption, vibration characteristics, firmware levels, revision levels, cooling requirements, actual performance, and quality. Network Appliance tests to ensure those specific characteristics working well together. So, just buying a 9 GB drive is not the answer. Buying to our engineering and purchasing specifications is what is needed. 4. Total Solutions. Network Appliance provides a total product solution. These drives are part of the total Network Appliance solution. Our total solution is not designed to work with other components, even if the raw component manufacturer is the same. 5. Support, Support, Support. Support and service are an integral part of our products. The advance replacement feature and technical support on our products is part of what sets drives purchased from Network Appliance apart from drives purchased from others. We stand behind our drives in a way other outlets won't. 6. Extensive quality control. NetApp has an extensive quality control system in place. We have to be sure of the whole process by which components (such as disk drive) are manufactured, tested, and delivered. Purchasing from another source prevents Network Appliance from ensuring the customer of the quality they depend on. Plus, the unique changes our vendors perform for us will not be found in drives purchased from other sources. 7. We fail what commodity outlets buy. Finally, drives that customers can buy from commodity outlets have failed the exacting quality, performance, and reliability standards of Network Appliance. Network Appliance has specifically rejected the drives that show up at these venues. In addition, drives at commodity outlets are subject to damaging handling and failure-inducing shock and vibration. Customers who buy from commodity outlets suffer greater data loss, shorter drive life, and higher incidences of failure and downtime. Regarding the question about using 9 GB drives in the same shelves as 4 GB drives, there are several guidelines that need to be heeded. First, the disk shelves that supported 4 Gb drives did not support properly the 9 Gb drives and network Appliance was forced to go to a new shelf design. Therefore, our engineering analysis determined that putting 9 GB drives in the 4 GB shelves would result in insufficient power and cooling to ensure the reliability our users have come to expect from Network Appliance. Second, there are guidelines for using 4 GB drives in new filer models. These guidelines take into account the operating system limitations of each filer. Be sure to review the technical product bulletins to ensure that the operating system limits on your filer are accounted for in terms of total storage capacity as well as the maximum number of slots available for the dual-channel SCSI adapters. Let me summarize these guidelines for you: Filer Storage Options F220 14 x 4 GB Fast Narrow SCSI drives F330 28 x 4 GB Fast Narrow SCSI drive OR 14 x 4 GB Fast Narrow SCSI and 7 x 9 GB drives F540 52 x 4 GB Fast Wide SCSI drives OR 26 x 4 GB Fast Wide SCSI and 13 x 9 GB Fast Wide SCSI drives Note: 4 GB count could include up to 24 Fast Narrow SCSI drives F210 14 x 4 GB Fast Wide SCSI drives OR 7 x 9 GB Fast Wide SCSI drives F230 14 x 4 GB Fast Wide SCSI drives OR 14 x 9 GB Fast Wide SCSI drives OR 7 x 4 GB and 7 x 9 GB drives F520 28 x 4 GB Fast Wide SCSI drives OR 28 x 9 GB Fast Wide SCSI drives OR 56 x 9 GB FC-AL drives OR 14 x 4 Fast Wide SCSI GB and 14 x 9 Fast Wide SCSI GB OR 21 x 4 GB Fast Wide SCSI and 7 x 9 GB Fast Wide SCSI OR 7 x 4 GB Fast Wide SCSI and 21 x 9 GB Fast Wide SCSI F630 52 x 4 GB Fast Wide SCSI drives OR 52 x 9 GB Fast Wide SCSI drives OR Up to 168 x 9 GB FC-AL drives OR Up to 84 x 18 GB FC-AL drives

2 1

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