toasters

toasters@lists.teaparty.net

13520 discussions

FW: crossover gigabit
by Francett, Robert D (SAIC) 29 Oct '99

29 Oct '99

fyi... > ---------- > From: Francett, Robert D (SAIC) > Sent: Friday, October 29, 1999 10:55 AM > To: 'Graham C. Knight' > Subject: RE: crossover gigabit > > Graham, > > We had to turn-on auto negotiate on our filer for the Gb connect to work. > This in the rc file too to ensure @ reboot. > #Turn-on auto negotiate > rc_toggle_basic > alttune 0 LinkNegotiation 1 > rc_toggle_basic > # > Regards, > Bob Francett > BP Exploration (Alaska) > > ---------- > From: Graham C. Knight[SMTP:grahamk@ast.lmco.com] > Sent: Friday, October 29, 1999 6:54 AM > To: toasters(a)mathworks.com > Subject: crossover gigabit > > > Does anyone know of any cheat sheets for setting up > a dedicated crossover gigabit ethernet between a filer > and an HP-UX system? > > I've gotten as far as a good link light, but cannot get > a ping to work. > > Thx, > Graham > >

1 0

crossover gigabit
by Graham C. Knight 29 Oct '99

29 Oct '99

Does anyone know of any cheat sheets for setting up a dedicated crossover gigabit ethernet between a filer and an HP-UX system? I've gotten as far as a good link light, but cannot get a ping to work. Thx, Graham

1 0

Is there a way to backup a specific snapshot...?
by Eyal Traitel 28 Oct '99

28 Oct '99

Hi! Does anybody know of a way to backup a specific snapshot using NDMP ? We're using 5.3.2 and up, Veritas NetBackup 3.11 and 3.2 Does anybody have experience with backuping Atria ClearCase ? I know there's a NOW article on that... Eyal.

2 1

Re: General VIF questions
by Sam Cramer 26 Oct '99

26 Oct '99

tkaczma> This is because I think that you cannot use IP addreses with trunk tkaczma> interfaces. At least you couldn't do it last time I set up tkaczma> "multi" trunking on a new rev of DOT. Right --- for vifs, partners must be specified by interface name, not IP address. Sam

1 0

Re: General VIF questions
by Sam Cramer 26 Oct '99

26 Oct '99

cramer> Which interface takes over is specified using the "ifconfig" cramer> "partner" clause. Please note that vif partners MUST be specified cramer> using the interface name (e.g. vif0), not the interface address cramer> (e.g. 172.27.5.50). This was not clearly spelled out in the cramer> documentation, and it has tripped up a number of people. Brian> Hrm, I'm not sure what you mean by using the interface address Brian> instead of the IP address. If I just give the partner the VIF name, Brian> how does it know what IP to use? Is that stored on the mailbox Brian> disk? When a node takes over for its partner, it reads the partner's /etc/rc file. The ifconfig commands are processed specially: they are examined for a "partner" clause. If a partner clause is found, the interface named in the partner clause takes over for the interface specified in the ifconfig command. If no partner clause is found, the ifconfig command is ignored. Sam

3 2

General VIF questions
by Brian Tao 26 Oct '99

26 Oct '99

I have two clustered pairs of F740's arriving soon, so I will likely be able to discover answers to most of my questions, but I figured I'd do a bit of legwork and check my assumptions while I'm still in the planning stages. I'll consider just one of the clustered pairs: nfs1 and nfs2. Both will be equipped with two QFE NIC's. They will connect to a pair of switches with an uplink between them (i.e., passing traffic between two hosts on the same subnet, but on different switches). Logically, consider it a single switch. Each Netapp will be visible on two subnets: 10.35.8.0/25 and 10.35.8.128/25. NFS clients in one subnet will not be able to see filer interfaces in the other. I want to create two single-mode VIF's on each filer, using four interfaces on each filer: ====Active VIF_Subnet1====[e1a] [e2a]----Standby VIF_Subnet1--- ----Standby VIF_Subnet2---[e1b] [e2b]====Active VIF_Subnet2==== ..........Unused..........[e1c] [e2c]..........Unused.......... ..........Unused..........[e1d] [e2d]..........Unused.......... Trunks are created across NIC's. Ports on e1 are cabled to switch 1, and ports on e2 are cabled to switch 2. Since this is a single- mode trunk, the switches shouldn't care, right? During normal operation, switch 1 only sees an active port from e1a, and switch 2 only sees an active port from e2b (since e2a and e1b are effectively "downed" interfaces). While it would be a bit less confusing to say "all interfaces on e1 are active, and all interfaces on e2 are on standby", I prefer to have an active interface on each NIC at all times. This should lessen the chance of Murphy's Law kicking in on an early Sunday morning when e1 does fail, only to reveal that e2 had also died quietly sometime before then. I also want those same physical interfaces to assume their partner's VIF configuration in a failover situation (i.e., what the manual calls "shared interface takover"). So this is what I have, based on the 5.3.2 documentation: nfs1> vif create single VIF_Subnet1 e1a e2a nfs1> vif create single VIF_Subnet2 e2b e1b nfs1> vif favor e1a nfs1> vif favor e2b nfs1> ifconfig VIF_Subnet1 10.35.8.1 netmask 255.255.255.128 partner 10.35.8.2 nfs1> ifconfig VIF_Subnet2 10.35.8.129 netmask 255.255.255.128 partner 10.35.8.130 nfs2> vif create single VIF_Subnet1 e1a e2a nfs2> vif create single VIF_Subnet2 e2b e1b nfs2> vif favor e1a nfs2> vif favor e2b nfs2> ifconfig VIF_Subnet1 10.35.8.2 netmask 255.255.255.128 partner 10.35.8.1 nfs2> ifconfig VIF_Subnet2 10.35.8.130 netmask 255.255.255.128 partner 10.35.8.129 This configuration should be able to withstand failure of a filer NIC port, a filer NIC, an Ethernet cable, a switch port, an entire switch and an entire filer. Certain combinations of double failures are also non-lethal. Since this is all theoretical (to me, anyway), I have some questions. :) What characters are allowed in a VIF name? Is anyone using single-mode VIF's across two different switches like this? How does the Netapp know which VIF to use with the "vif favor" command? I assume this means a physical interface can only participate in one VIF at a time. Is it possible to prioritize interfaces in a VIF with more than two ports trunked together? Can VIF's be shared in a cluster? Can they share ports as above? Special mention is only made about super VIF's and clustering in the 5.3.2 SAG. Can super VIF's be created from single-mode VIF's, or only multi-mode VIF's? Again, the manual only mentions the latter, but does not prohibit the former. Can super VIF's themselves be multi-mode, or single-mode only? I'll post the results of my configuration testing once I have the hardware to play with. -- Brian Tao (BT300, taob(a)risc.org) "Though this be madness, yet there is method in't"

2 3

dlt8000 supported?
by Luke Mewburn 26 Oct '99

26 Oct '99

Hi All. Is the DLT-8000 supported in OnTap? I've been quoted a price on a DLT-8000 that is A$100 more than a DLT-7000, and when you're paying nearly A$8000 for a tape drive that seems like a good deal to me. Hopefully a DLT-8000 should act like a DLT-7000 given the same tape density SCSI control codes. I'd rather spend a little bit more money now for a drive with 12.5% more capacity, even if I have to wait for a newer release to get that extra density. Luke.

1 0

Re: mailboxes on toaster
by Jim McCoy 26 Oct '99

26 Oct '99

George Kahler wrote: >Q: Will mbx buy me performance on the toaster ? >Q: Is Maildir a better choice ? As in all difficult questions the answer is "it depends." People have given some very good reasons on why maildir is an appropriate choice given certain constraints, so let me mention a few things that favor mbox-format mail storage on a filer. We run several tens of millions of user mailboxes on a fleet of NetApps for YahooMail and so here are a few other points to consider that we have run into over the past few years: 1) Locking is a dead issue for mbox. There is no way to adequately provide file locking via the standard channels so you will probably need to deal with the locking issue as a seperate problem. We ended up writing our own mailbox locking system but you might be able to use procmail for as your local delivery agent and let that try to figure out the locking issues. Locking is important, but do not fixate on this issue and ignore all of the other points that you need to consider. 2) Inode cost should not be overlooked. You get at most 32 million per volume, so if you have a lot of mailboxes a maildir or other one-file-per-message scheme will chew through your inode limit long before you run out of disk space on that volume. To compensate you will have to make smaller volumes. This is not necessarily a bad thing but each volume will be at least one additional raid group and on additional wasted parity disk (increased cost/mailbox). The larger number of inodes in use will also increase the time it takes your filer to complete a wack run and will have an impact on backup speeds. 3) Mail messages are small. There is a simple bimodal distribution to internet mail message size, they are either very small or very large. The small files will cost you storage bits because the filer uses 4K to store each 500byte message. The small file sizes will also keep your read chain length very low so we will get significantly decreased performance on your backups of this volume and you will notice some decreased read performance on the filer in general. jim

4 7

Re: mailboxes on toaster
by jrg＠gbnet.net 25 Oct '99

25 Oct '99

On Mon 25 Oct, 1999, "Luke Gain" <luke(a)erinet.com> wrote: >one thing I found to be a big win is to make sure all the filenames are 12 >characters >or less. evidentally, the netapp only caches names that are <12 characters >long. >so if the name is >12 characters it takes a disk-op to read the directory. I expect it'll interact with your operating system, too (or this may actually be the sole cause.) Eg. Solaris puts namei lookup results in the DNLC (directory name lookup cache) only if they're under a certain size (vmstat -s shows these as "toolong"), so if you've got long directory paths you would expect to see a problem not dissimilar to the one you describe. On the subject of the DNLC, beware of making it too big as that has been bad for performance too. ("classic" BSD sets this "toolong" constant NCHNAMLEN at 31. FreeBSD-current seems to have done away with it.) James.

1 0

Re: General VIF questions
by Sam Cramer 25 Oct '99

25 Oct '99

>>>>> "Brian" == Brian Tao <taob(a)risc.org> writes: Brian> Since this is all theoretical (to me, anyway), I have some Brian> questions. :) Brian> What characters are allowed in a VIF name? A vif name may be composed of any non-blank characters and should begin with a letter. The name may be at most 15 characters long. Brian> Is anyone using single-mode VIF's across two different switches like Brian> this? Yes. We have a number of customers who use single-mode vifs composed of links attached to two different switches to protect against switch failure. Brian> How does the Netapp know which VIF to use with the "vif favor" Brian> command? I assume this means a physical interface can only Brian> participate in one VIF at a time. Yes, an interface (physical or virtual) can only participate in one vif at a time. Brian> Is it possible to prioritize interfaces in a VIF with more than two Brian> ports trunked together? If you're asking "can you favor a link in a single-mode VIF composed of more than 2 links", the answer is yes. If you're asking "can I specify secondary, tertiary (and so on) priority for links in a single-mode VIF", the answer is no: only one link may be favored; all others are equally "non-favored". Brian> Can VIF's be shared in a cluster? Can they share ports as above? Brian> Special mention is only made about super VIF's and clustering in the Brian> 5.3.2 SAG. vifs can't be shared across a cluster. However, the surviving node may takeover for a vif just like it may takeover for other interfaces. Which interface takes over is specified using the "ifconfig" "partner" clause. Please note that vif partners MUST be specified using the interface name (e.g. vif0), not the interface address (e.g. 172.27.5.50). This was not clearly spelled out in the documentation, and it has tripped up a number of people. Brian> Can super VIF's be created from single-mode VIF's, or only Brian> multi-mode VIF's? Again, the manual only mentions the latter, but Brian> does not prohibit the former. Super vifs may be created from single-mode vifs or multi-mode vifs. I'm not sure why you would want to create a super-vif from single-mode vifs, though. Brian> Can super VIF's themselves be multi-mode, or single-mode only? Single mode only. Sam

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