toasters April 2001

toasters@lists.teaparty.net

205 participants
286 discussions

Start a nNew thread

REMOVE
by Ching Chuan Kuo 24 Apr '01

24 Apr '01

REMOVE

1 0

RE: data migration
by Sims, Aaron 23 Apr '01

23 Apr '01

ndmpcopy is the best data migration method based on the situation you described. ndmpcopy will have to be run from a Unix system. This is basically the migration controller. You can do a rdist style refresh (like a incremental backup), just use the -level 0 option for the initial copy, then you can do -level 1 (up to 9) for your incrementals. Best method is to setup crossover GbE between the filers. HTH, Aaron > -----Original Message----- > From: Leonard, Roger [mailto:Roger.Leonard@marconi.com] > Sent: Monday, April 23, 2001 3:18 PM > To: 'toasters(a)mathworks.com' > Subject: data migration > > > i have 4 filers (330's and a 540) that i need to migrate to > one 840. the > 840 is brand new with lots of empty 72GB disks and space is > not an issue. > we were initially going to use snapmirror, however the old > machines have > volumes of varying sizes, from 120GB on up to 400+. with the > 72gb drives, > we would have some inefficent raid groups if we did a > volume-volume copy. > plus i want to shuffle and shift the data around during this move. a > "refreshable" copy at the qtree or directory level from old > filer to the new > is what we need. it looks like ndmpcopy will support this. > has anyone > ever used this? is this something we can do from the filer > CLI?? i didn't > see anything at the help prompt. maybe we need to download > ndmpcopy and do > it from solaris after starting ndmpd on the filers? my > downtime would be > minimal as well, so a quickly refreshable (ala rdist) copy is > needed. any > thoughts would be appreciated. > > thanks > > Roger D. Leonard >

1 0

data migration
by Leonard, Roger 23 Apr '01

23 Apr '01

i have 4 filers (330's and a 540) that i need to migrate to one 840. the 840 is brand new with lots of empty 72GB disks and space is not an issue. we were initially going to use snapmirror, however the old machines have volumes of varying sizes, from 120GB on up to 400+. with the 72gb drives, we would have some inefficent raid groups if we did a volume-volume copy. plus i want to shuffle and shift the data around during this move. a "refreshable" copy at the qtree or directory level from old filer to the new is what we need. it looks like ndmpcopy will support this. has anyone ever used this? is this something we can do from the filer CLI?? i didn't see anything at the help prompt. maybe we need to download ndmpcopy and do it from solaris after starting ndmpd on the filers? my downtime would be minimal as well, so a quickly refreshable (ala rdist) copy is needed. any thoughts would be appreciated. thanks Roger D. Leonard

1 0

OT - swag
by Chuck Ross 23 Apr '01

23 Apr '01

Man, you guys suck! When I went to Net App training all we got was a travel coffee mug and a really lame tote bag. Chuck Ross > > > I think I've got everyone on this thread beat. I am the > proud owner of a > black leather NetApp jacket. Currently the best vendor swag > in my whole > wardrobe. > > -- Jeff > > > On Fri, Apr 20, 2001 at 02:54:44PM +0200, Leif Nixon wrote: > > "Timothy Cook" <TCOOK(a)joy.com> writes: > > > > > Maybe I'm missing out on the really important aspect of netapps??? > > > > > > Other than a jacket I got at Admin training I've never received my > > > new clothing line from Netapp! Is this an optional part of the > > > Support Agreement, or should I start complaining that I'm not > > > receiving the level of support from Netapp that I should be! > > > > Yes, you might want to consider that. > > > > A couple of months ago I added a new 36-gigger to my F720 setup, and > > since I failed to notice the little sticker on the bottom > of the disk > > about supported ONTAP versions, the filer halted and production > > stopped. Ouch. > > > > Not two hours later I received an overseas FedEx package containing > > a NetApp sweater. Now that's service! > > > > -- > > Leif Nixon Sysadm/developer Ericsson SoftLab AB > > ---------------------------------------------------------- > > E-mail: nixon(a)softlab.ericsson.se Phone: +46 13 23 57 61 > > ---------------------------------------------------------- > > >

5 4

Re: How would you respond to this -Reply -Reply
by Timothy Cook 23 Apr '01

23 Apr '01

Humorously enough, since I posted my response to Chris, I have actually received 2 shirts. One came from a vendor (who monitors this forum) whose software I am presently evaluating (thank you Donna and Judy), and the other came from ..... Well, perhaps I had better not say. Suffuce it to say that the next time a Netapp sales rep comes around he may discover that some of his competitors are not so cheap with the swag! >>> "Timothy Cook" <TCOOK(a)joy.com> 04/20/01 08:35am >>> Maybe I'm missing out on the really important aspect of netapps??? Other than a jacket I got at Admin training I've never received my new clothing line from Netapp! Is this an optional part of the Support Agreement, or should I start complaining that I'm not receiving the level of support from Netapp that I should be! For the benefit of you netapp guys who monitor this board, I wear an Extra Large. Of course, in Netapps defense, I've never received any EMC shirts either. (p.s. don't bother with the boxer shorts, I won't wear them) >>> Chris Lamb <skeezics(a)measurecast.com> 04/19/01 07:00pm >>> My story about EMC is rather different, I guess. Some local EMC reps scheduled a visit with my boss and came to visit us a while back. I wore not one but _two_ NetApp shirts that day - the very cool engineering shirt (black pre-F630 shirt with Mothra on the back) and the snazzy denim Land's End shirt with the embroidered logo on it. The night before I printed a copy of the recent INPUT survey, highlighting the part where I'm quoted (anonymously), and left it on my boss's chair. He brought it to the meeting. They arrived. I smiled and shook hands and did the business card thing. We sat down at the conference table. They asked their questions, and I politely answered them. It went something like this: "What about reliability?" they asked. "I've been running NetApps since 1995, and to the best of my recollection I have never once lost data due to hardware or software failure," I replied. "Yes, but what about uptime?" they inquired. "alexandria> uptime 3:32pm up 187 days, 14:10 1162740653 NFS ops, 129052586 CIFS ops, 0 HTTP ops colossus> uptime 3:32pm up 171 days, 8:48 1692312427 NFS ops, 0 CIFS ops, 0 HTTP ops", was my reply. "Expandability?" they suggested. "I have 21 free slots for more disks - I bought the box with extra shelves and a 2nd controller so that I could grow it on-line without any need for downtime. I can grow our box to > 1TB without even thinking hard." "Performance!" they pleaded. "My Oracle DBA is happy, and says he can't see a difference between the 760 and a locally-attached dual-loop A5200. Besides, with a head-swap we can trade up to the 800-series, or cluster, or trade-up and then cluster, and I've got all the slots I need to add extra Gigabit cards..." "Yes, but our 'phone home' support feature and customer service are second to none!" they intoned. "Actually," I said, "your story about the hardware tech showing up on a Saturday with a replacement disk for a demo machine reminds me of a similar incident... one day a box arrived unexpectedly from Netapp with a replacement drive in it, which puzzled us because we hadn't reported a failure. But sure enough, we walked into the server room and one of the disk canisters had a red light on. Five minutes later we swapped it out and that was that. Sure a lot cheaper than sending a field circus guy out to do a job that any basic systems operator can do." (Ouch, that stung a bit, but hey, I _like_ playing with the hardware, and if I have one complaint about the reliability of my filers it's that after I bolt it into the rack and plug it in, I don't get to fool around with it much. That's a blessing, of course. :-) "And as for service, a funny thing happened in the middle of the night a few months back, when I was innocently rearranging some power cords to balance the load on a split-phase UPS, oblivious to the fact that the 'autosupport' emails were being sent out. Imagine my surprise when an email arrived 20 minutes later from the Netherlands, asking if I needed to have a replacement power supply shipped..." They got the point. Clearly, they weren't going to make a sale that day. :-) But we listened to their presentation, and had a pleasant enough chat, and that was that. I had done my homework about their product line, too, and I think they knew it. They didn't try to BS me, and they didn't try to badmouth Netapp. So I can't comment on whether or not this is typical behavior from EMC reps, but my experience wasn't at all negative. Mostly, though, it's the shirts. If my vendors didn't clothe me I'd have nothing to wear! :-) -- Chris (still waiting for NetApp boxer shorts) -- Chris Lamb, Unix Guy MeasureCast, Inc. 503-241-1469 x247 <skeezics(a)measurecast.com>

1 0

Re: Vif and Cluster setup questions
by Sam Cramer 23 Apr '01

23 Apr '01

"FPS France" <fpsfrance(a)fps.fr> writes: > Also , I wonder if you do not bad in using identical names for your > devices (vif01 in both Filer). This shouldn't cause a problem. Using identical vif names on each node in a cluster is common, and I'm not aware of any bug arising from it. Sam

1 0

Re: How would you respond to this -Reply
by Colm Buckley 23 Apr '01

23 Apr '01

> Hey! You guys got a knife? Geez. We should start up a list of stuff > sorted by gee-wow value so that we can each find out where we stand > with Net App... One grey T-shirt. (Thanks, Matt!) Even though I've adopted a policy of "all of my employers are belong to NetApp" - can't be without my trusty toasters, wherever I go... okay, admittedly my employers can only generally go for things like single F720s, but it's the *principle* that counts. Trade shows and expos really don't have the quality of swag these days; last year's "eXpo" in Dublin (software centre of Europe, and all that) netted me one T-shirt (thanks again, Matt!), a couple of pens, and heaps of squeezy rubber desk toys (just missed out on the Corel inflatable Tux penguins). Even five years ago, you could count on a couple of Microsoft t-shirts (for defacing as appropriate) and the occasional Cool Inflatable Thing. Ah well. Colm (*sigh*) -- Colm Buckley, NewWorld Commerce, 16 South Cumberland St., Dublin 2, Ireland. Personal: +353 87 2469146 / colm(a)tuatha.org / http://www.tuatha.org/~colm/ Business: +353 1 4334334 / colm(a)nwcgroup.com / http://www.nwcgroup.com/ I'm already visualizing the duct tape over your mouth.

1 0

Re: Vif and Cluster setup questions
by Sam Cramer 23 Apr '01

23 Apr '01

Art Hebert <art(a)arzoon.com> writes: > 2 F760 filers clustered together, onboard ethernet (e0), 10/100 nic (e6) > Cisco 6509 switch trunked together > > I'm trying to get the e0 and e6 ports to work failover for each other and > ideally load balance to get better performance. A multi-mode vif will, when attached to a switch or set of interfaces on another host capable of trunking, balance network load across its constituent links. Most switches distribute the load across the links by hashing an incoming packet's source and destination address. Thus, if incoming packets share the same source and destination address, they will flow over the same link. This may be what is biting you here. Of course, you need to configure the switch to tell it that the switch ports you connected to the filer's e0 and e6 ports comprise a trunk (or, to use Cisco terminology, an EtherChannel). If you don't do this, traffic will also flow over only one link. > I've been able to get the virtual interface running and fail the cluster > and everything is fine but I can't get it to load balance (I even tried 4 > clients hitting it copying 500mb files) and it seems to always want to > use e0, even if I do a "vif favor e6" "vif favor" is for use with single mode vifs only. I don't think this is your problem, but you don't need (and shouldn't use) "vif favor" in this configuration. > Here is my rc lines: > > ifconfig e6 mediatype 100tx-fd > ifconfig e0 mediatype 100tx-fd > vif create multi vif01 e6 e0 > vif favor e6 > ifconfig vif01 `hostname`-e6 mediatype 100tx-fd partner vif01 ^^^^^ When specifying the partner of a vif, use the IP address of the partner interface, not the interface name. > Does it matter what order the e6 and e0 commands come in, seems that > it affects how vif stat reports? Order shouldn't matter, save for statistics output. > Can I even use e0 and e6 together or should I have another card? its on > the 32 bit bus and performance is half of e6. The vif logic doesn't care. Whether the performance will be acceptable is your call. Sam

2 1

Flash Shose!!��Flash Shose!!��
by u417787520＠yahoo.com 22 Apr '01

22 Apr '01

1 0

Flaying Shose!!��Flaying Shose!!��
by x9334400＠hotmail.com 21 Apr '01

21 Apr '01

1 0

← Newer
1
...
8
9
10
11
12
13
14
...
29
Older →

Jump to page:

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 April 2001