toasters August 1999

toasters@lists.teaparty.net

104 participants
129 discussions

Re: [Fwd: disk_fw_update]
by raster＠netapp.com 05 Aug '99

05 Aug '99

>I was reading the docs prior to upgrading to 5.2.3, and I realized that I get >to upgrade everything, os, filer fw, and disk fw.. > >The thing that occured to me as unfortunate was from na_disk_fw_update(1): > > This command makes disks inaccessible for up to 2 minutes, so network > sessions using the filer should be closed down before running it. > >I've only got 4 shelves, so its not a huge deal, but I could see this having a >big impact in larger environments.. > >Obviously the disk has to be inaccesible during the period of the fw upgrade, >but disabling file service necessary? The reason I ask is that the filer can >already run short a disk following a failure, so why not copy this >functionality so that disk_fw updates can be done without service disruption? Hi Kevin. You are correct that a RAID system protects against a single disk failure; the raid group is placed into degraded mode and the contents rebuilt onto a spare. Now, to address your point, yes, it's technically possible to place a raid group into "pseudo-degraded" mode where a disk is temporarily take offline but no reconstruction is started. Note, OnTap doesn't work this way and making it work this way would take quite a bit of doing. For one thing, "re-syncing" the disk which was temporarily taken offline and then brought back later would take quite a bit of book keeping and associated headaches. *If* the above machinery was in place, firmware updating could be made to work one-disk-at-a-time without any disruption of service. However, one has to ask oneself whether it's worth it? How often does one upgrade disk firmware? Hope this answers your question. Let me know otherwise. -- --------------------------------------------------------------------------- Radek Aster [ raster_at_netapp_dot_com ] | "I bike, therefore I am" http://www.nowhere.net/~raster/ | - Rene' DesBiketes ---------------------------------------------------------------------------

1 0

RE: Adding a FC shelf in a cluster w/o downtime
by Norman, Paul 05 Aug '99

05 Aug '99

Actually it is documented in the "Fibre Channel StorageShelf Hardware Guide". There is a chapter on doing exactly this although the doc says attach A channel first. You can see it on-line on NOW: http://now.netapp.com/knowledge/docs/hardware/fcshelf.pdf Its in appendix A, entitled "Adding Hardware for a Cluster". I'm not surprised that you didn't find it. I'm the product manager for Clusters and I've reviewed every manual numerous times. I knew it existed somewhere but it still took me 15 minutes to dig it up. Unless your one of those people who reads every manual cover to cover you probably would never have known it existed. The Tech Pubs crew here has done major reorganizations of the manual set. They have put all the software manuals on line in HTML in addition to the pdf format. They are always looking for ways to improve the doc set and welcome comments from customers on things they can do to make the docs easier to use (It would be nice if you told them when they've done something you like too, but that might be pushing it). You can send comments via feedback on NOW or send e-mail to doccomments(a)netapp.com. When you run across issues like this, where you can't find something, a quick note saying what you wanted and where you looked might help the next time they reorganize the manual to put the information in a better place or create a better index or whatever. PS. In house we just rack the shelf and shut down, add the daisy chains, power up. It only takes a five minutes, but then we're pretty adept at cabling since we get so much practice, YMMV. Cheers Paul Norman, Product Manager Direct: (408) 822-6288 Network Appliance E-Mail: pnorman(a)netapp.com 495 East Java Drive Corp: (408) 822-6000 Sunnyvale, CA 94089 Web: WWW.NetApp.Com > -----Original Message----- > From: Mohler, Jeff [mailto:jeff.mohler@wilcom.com] > Sent: Thursday, August 05, 1999 9:42 AM > To: Jay Soffian; toasters(a)mathworks.com > Subject: RE: Adding a FC shelf in a cluster w/o downtime > > > My initial feelings are 'bad things will happen'. > > I dont have any technical data to back that up, other than its not > documented to do it that way. > > -----Original Message----- > From: Jay Soffian [mailto:jay@cimedia.com] > Sent: Thursday, August 05, 1999 10:05 AM > To: toasters(a)mathworks.com > Subject: Adding a FC shelf in a cluster w/o downtime > > > > We are going to be adding an additional FC shelf to our clustered > F740's. Obviously, the safe way to do this is to halt both 740's, > attach the A & B loops, then bring both filers back up. > > But, what I was wonder was, would it be possible to: > > - Attach the B-loop first. > - Have the B-Loop Filer take over all th shelves (including > the new shelf). > - Attach the A-Loop. > - Give back the shelves to the A-Loop filer. > > I don't think we'll do this as we're only talking about about maybe 2 > minutes of downtime vs 30 seconds, but I was wondering if this is a > feasible thing to do. > > Thoughts? > > j. >

2 1

Adding a FC shelf in a cluster w/o downtime
by Jay Soffian 05 Aug '99

05 Aug '99

We are going to be adding an additional FC shelf to our clustered F740's. Obviously, the safe way to do this is to halt both 740's, attach the A & B loops, then bring both filers back up. But, what I was wonder was, would it be possible to: - Attach the B-loop first. - Have the B-Loop Filer take over all th shelves (including the new shelf). - Attach the A-Loop. - Give back the shelves to the A-Loop filer. I don't think we'll do this as we're only talking about about maybe 2 minutes of downtime vs 30 seconds, but I was wondering if this is a feasible thing to do. Thoughts? j.

1 1

managing permissions in mixed qtrees
by Gregory M. Paris 05 Aug '99

05 Aug '99

Toasters, We have a new filer with NFS and CIFS. The plan is to move/merge all of the NT and UNIX homes (from NT and UNIX servers) onto mixed qtrees. Each user gets one and only one home directory. Conceptually simple. However, we now see one thing that may bite us in practice: Doing recursive mode/ACL changes in user home directories will have the effect of converting some UNIX-security files to NT-security files or vice versa (depending on whether the operation done from NT or UNIX). At a minimum this will cause confusion, but in some cases could break things -- most probably things that depend on NT ACLs that get wiped when somebody does chmod. I'd like to hear from other admins that have merged UNIX/NT home directories. Have you had this kind of problem? If so, have you found ways to work around it? Thanks i.a., Greg -- Gregory M. Paris CIS Engineering Systems Bose Corporation Email: paris(a)bose.com The Mountain Phone: +1 508 766-6080 Framingham, MA 01701-9168 FAX: +1 508 820-0173

1 0

RE: NetApp/NDMP
by Melvin, Grant 05 Aug '99

05 Aug '99

Hi Bruce, I still foresee a role for NDMP in a hybrid NAS/SAN solution: 1. Scheduling backups 2. Maintaining catalog of backed up images 3. Changing tapes within a library Cheers, Grant -----Original Message----- From: sirbruce(a)ix.netcom.com [mailto:sirbruce@ix.netcom.com] Sent: Wednesday, August 04, 1999 9:47 PM To: michael.marchi(a)netapp.com Cc: toasters(a)mathworks.com Subject: Re: NetApp/NDMP On 08/04/99 15:27:17 you wrote: > NDMP is the strategic direction for backup at > Network Appliance. We see NDMP as a winner for Network > Attached Storage (NAS) vendors and ultimately we > think this could also be very prevalent in Storage > Area Network (SAN) environments too. When Netapp has a hybrid NAS/SAN solution, I would expect straight "local" backup to a tape device on the SAN to be an option, via traditional dump, without NDMP involved. Bruce

1 0

tape drives in a cluster
by kevin graham 05 Aug '99

05 Aug '99

Does anyone know if there's a way to modify the scsi id of the tape controller in a 760? It would be nice if our clustered 760's could share a tape drive, or pair of drives, especially given how there's a whole controller set aside for it.. ...kg..

1 0

ok, so what do you use?
by dnb＠ccs.neu.edu 05 Aug '99

05 Aug '99

Howdy- The whole backup discussion leads me to a (potentially loaded) question: what software does NetApp (the company) use to back up its data? Peace, dNb

1 0

RE: Adding a FC shelf in a cluster w/o downtime
by Mohler, Jeff 05 Aug '99

05 Aug '99

My initial feelings are 'bad things will happen'. I dont have any technical data to back that up, other than its not documented to do it that way. -----Original Message----- From: Jay Soffian [mailto:jay@cimedia.com] Sent: Thursday, August 05, 1999 10:05 AM To: toasters(a)mathworks.com Subject: Adding a FC shelf in a cluster w/o downtime We are going to be adding an additional FC shelf to our clustered F740's. Obviously, the safe way to do this is to halt both 740's, attach the A & B loops, then bring both filers back up. But, what I was wonder was, would it be possible to: - Attach the B-loop first. - Have the B-Loop Filer take over all th shelves (including the new shelf). - Attach the A-Loop. - Give back the shelves to the A-Loop filer. I don't think we'll do this as we're only talking about about maybe 2 minutes of downtime vs 30 seconds, but I was wondering if this is a feasible thing to do. Thoughts? j.

1 0

RE: what's a good NFS write speed in MB/sec?
by Mohler, Jeff 05 Aug '99

05 Aug '99

THis is an interesting problem for NetApp. I purchased several NetApps for use on Oracle Servers for the Oracle DB. In evaluating all the info I too had problems relating nfs ops/sec to a disk subsystem throughput. --- Thats because you cant. NFS operations are many things, file creation, removal, buffering, data, lookups, info..etc. Writing a 1Gb file wont create nearly the operations that creating 1GB of 10k files. So, I've now got 2 F740 systems I'm looking at and so far like a lot. In using a single gigabit connection to a F740 I can easily get sequential writes of around 15mb/sec. At this point the Netapp is running CPU utilization of around +80% utilization. I've seen peaks of close to 18mb/s where the cpu utilization is pegged on 100%. --- This is about where you'll peg out Ive seen. Until the box has some serious CPU dedicated to RAID stripe calculations, and much MUCH more NVRAM..the platform is artificially limited. And if I understand correctly, your NVRAM caching becomes even less effective when you cluster. These are not BAD points about the filer at all, just places where improvements are taking place.

1 0

RE: Adding a FC shelf in a cluster w/o downtime
by Watanabe, Steven 05 Aug '99

05 Aug '99

> From: Jay Soffian [mailto:jay@cimedia.com] > We are going to be adding an additional FC shelf to our clustered > F740's. Obviously, the safe way to do this is to halt both 740's, > attach the A & B loops, then bring both filers back up. > > But, what I was wonder was, would it be possible to: > > - Attach the B-loop first. > - Have the B-Loop Filer take over all th shelves (including > the new shelf). > - Attach the A-Loop. > - Give back the shelves to the A-Loop filer. > > I don't think we'll do this as we're only talking about about maybe 2 > minutes of downtime vs 30 seconds, but I was wondering if this is a > feasible thing to do. Adding a disk shelf requires the FCAL cable to be unplugged while that shelf is being rewired. This requires the owning node to be HALTed at the time. It is possible to keep one member of the cluster alive during shelf rewiring. The approach is to HALT one node, failover to the other node, wire up the first node's A loop, and then bring that first node back to life. Then repeat the process, failing over and rewiring in the opposite direction.

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 August 1999