toasters

toasters@lists.teaparty.net

13513 discussions

new quota tools: netapp-qdiff and netapp-qformat
by Daniel Quinlan 25 Mar '98

25 Mar '98

I cleaned up two more small scripts of mine sufficiently that I don't mind showing them to other people. The first is "netapp-qdiff", which essentially subtracts one quota report from another, and shows you what changed between the two. It only works on NetApp 5.0 right now, but if 4.x capability is wanted, I'll consider trying to add it. The second is "netapp-qformat", which prints out the quota report in a format that my small brain can read. It lines things up, converts to MB, prints decimal fractions for small usages, and sorts by tree name and disk usage. It should work for both NetApp 4.x and 5.0. The two scripts can work together, so here's what the output of a netapp-qdiff piped through netapp-qformat might look like. It's a little wide, just like the 5.0 quota report format: M-Bytes Files Type ID Volume Tree Used Limit Used Limit Quota Specifier ----- -------- -------- -------- ------- ------- -------- -------- ----------------- tree 1 vol0 local 0 0 2 - /local tree 2 vol0 transmeta-ht 10 0 57 - /transmeta-htdocs user joe vol0 transmeta-ht 8 0 10 - user bob vol0 transmeta-ht 2 0 47 - user mary vol0 transmeta-ht 0.1 0 4 - user jill vol0 transmeta-ht 0.1 0 3 - user billy vol0 transmeta-ht 0 0 -1 - user pat vol0 transmeta-ht 0 0 -2 - Also, instead of continuing to post my scripts here, I stuck the last three at <URL:http://www.pathname.com/~quinlan/netapp/>. So, is anyone finding this stuff useful or does everyone here administer their filers from Windows? - Dan

1 0

Re: splitting the load
by Daniel Quinlan 23 Mar '98

23 Mar '98

Tom Yates <madlists(a)mathworks.com> writes: > i'm planning to migrate about one-third of an existing F330's filespace > to another F330. what i want to do is move the most frequently accessed > third, so it has its own dedicated toaster. > > there are currently three file trees, each separately listed in exports, > each with its own CIFS share. is there any kind of monitoring tool i can > use to give me some idea of which of these three is the most heavily used? There is a program called nfswatch (a network snooper), but it only handles NFSv2 and may not have the information you want. If the traffic to each of the three file trees is used more by certain hosts, another method is to use the per-client statistics to make an educated guess. You could also monitor the turnover on each tree's filesystem - atime changes, mtime changes, etc. - Dan

1 0

Re: NetApp Support
by sirbruce＠ix.netcom.com 23 Mar '98

23 Mar '98

On 03/23/98 10:28:58 you wrote: >When I've got a bug that's affecting my customers and the quality of >service I'm offering, I could care less if its been "sanitized" and >internal engineering comments / cuss words / etc has been removed - I >want the *info available*, or at least indication that the problem >*is* an acknowledged bug and NetApp is working on it, as *soon as >possible*. Not to be an apologist for Netapp, but I have seen other reasons offered from time to time. For example, internal bug reports can contain portions of source code or other proprietary facts about the internal workings of the filer, bugs that are mischaracterized (which could mislead the unsuspecting customer into thinking it is something else), or even disagreements between engineers over the nature of the bug involved. I think these sorts of things should be filtered. Furthermore, some bugs get discovered in internal testing that do not get immediately fixed at release. Someone would have to, at release, identify these bugs that still exist and move them into the "public", as it were. >NetApp, if you're going to keep a list of bugs on your web page, make >sure they're up-to-date, otherwise such info is useless. >I've made this request at least twice, and never heard anything of it. I agree. I'm sure Netapp agrees as well. Putting the manpower and expertise together to do that, however, is not as easy as it may seem on a first pass. >The other "minor annoying" problem I've seen with the Toasters: > >nfs> exit >Use control-D to exit >nfs> ^D >Connection closed by foreign host. > >Why not just use the code that is in the OS to print "Use control-D to exit" >to make the filer go ahead and exit? I got an answer to this when I first asked it during the 1.x days. :) You may have noticed by now that the filer doesn't really support multiple tty sessions. If you're on the console and someone else telnets in, you're really multiplexed off the same character stream. So when the command parser gets an "exit", it doesn't know which connection(s) it came from. The control-D, on the other hand, is caught higher up in the input driver. I suppose one could check to see if there is only one session. Or there may be a way to put together some code to determine where the command came from. But the real fix would be to support multiple concurrent sessions. With the emphasis on GUI and rsh and such in more recent releases, that feature may not even be on the roadmap anymore. Or, maybe it's in 5.x? Bruce

2 1

splitting the load
by Tom Yates 23 Mar '98

23 Mar '98

i'm planning to migrate about one-third of an existing F330's filespace to another F330. what i want to do is move the most frequently accessed third, so it has its own dedicated toaster. there are currently three file trees, each separately listed in exports, each with its own CIFS share. is there any kind of monitoring tool i can use to give me some idea of which of these three is the most heavily used? Tom Yates - Unix Chap - The Mathworks, Inc. - +1 (508) 647 7561 MAG#65061 DoD#0135 AMA#461546 1024/CFDFDE39 0C E7 46 60 BB 96 87 05 04 BD FB F8 BB 20 C1 8C

1 0

Re: NIS Binding Problem
by Jason D. Kelleher 23 Mar '98

23 Mar '98

In message <35119FA5.91C737AD(a)lucent.com>, Dave Heiland writes: >This annoys me since I reported last year when I first set up my Netapp tha >t NIS >wasn't working right. Performance was awful, and it was rebinding to a >different NIS slave every couple of seconds. I never heard much back, and >because I immediately switched to using local netgroups I didn't have the >problem again. At least there will finally be a fix. > >Dave I wouldn't get your hopes up. I was sent that message over 5 months ago... jason >> When we first turned on NIS, everything went to hell. The filer >> started denying mounts to all the clients. Unfortunately, it didn't do >> this right away, took several hours. Seems the filer wasn't getting >> netgroup info. Anyway, here's what NetApp told me: >> >> > Hi Jason, >> > >> > I had one of our escalation engineers look over this >> > call with me. >> > >> > We believe that what you saw when upgrading to 4.1d >> > is the result of an inefficient method we use to >> > "lookup" the YP information. Instead of doing 3 or 4 >> > lookups to get the netgroup info when the mount is >> > checked we would do one for *every* line in the netgroup >> > file. >> > >> > This behavior has been logged as bug#4088 in our database >> > and engineering is working to modify the NIS client to >> > correct this problem. >> > >> > For now the recommendation would be to continue using >> > the local file method which hasn't caused you any >> > problems. When a fix is available for #4088 then you >> > could upgrade and begin using the NIS client for >> > netgroup which should be much faster. One of the >> > benefits of NIS client is that these lookups should >> > be faster but with our first release of this NIS >> > client they simply are not. >> > >> > I have flagged this call record as bug#4088 with all >> > the info you sent to me. The escalation engineer I >> > spoke with is actually the one that will fix the NIS >> > code and by flagging your call record as this bug >> > he can review it during the time he makes the patch. >> > >> > This would first be available as a patch release >> > (a release with a capital D in it such as you >> > were running prior to upgrading to 4.1d). The >> > best way to find out when it is available would be >> > to call back and reference the bug#4088 and ask >> > if Eirik has a patch for the NIS client yet. I >> > would think that you might want to wait 30 days or >> > so -- there is no schedule for a fix date for this >> > right now. >> > >> > Regards, >> > Gil >> >> I never bothered to check back. NOW lists bug #4088 as unfixed. >> Here's our nsswitch.conf which "solved" our problems: >> >> # >> # nfssrv1:/etc/nsswitch.conf >> # >> >> passwd: files nis >> group: files nis >> >> hosts: files nis dns >> netgroup: files nis >> >> Good Luck. >> >> --- >> Jason D. Kelleher kelleher(a)susq.com >> Susquehanna Investment Group 610.617.2721 (voice) >> 401 City Line Ave, Suite 220 610.617.2916 (fax) >> Bala Cynwyd, PA 19004-1122

7 10

Re: netapp-top - display current top NFS clients for a NetApp
by sirbruce＠ix.netcom.com 23 Mar '98

23 Mar '98

On 03/23/98 09:36:38 you wrote: > >Nice little script. Now, the question is, what sort of performance hit will >I take if I turn on nfs.per_client_stats.enable ? I am assuming that on a >moderately loaded f630 it shouldn't be much of a problem. Very little. I know it was measured at one point but it was nothing very important. Probably less than the impact of quotas. And more than worth the tradeoff. For an F630 I doubt it would even register a 1% difference. Bruce

1 0

Re: NetApp Support
by Andrew Hobson 23 Mar '98

23 Mar '98

On Mon, 23 Mar 1998 09:23:00 -0500, Gerard Hickey <Gerard.Hickey(a)nsc.com> said: > Does anyone else out there have problems with NetApp support? Or is it > just me? We've had some serious support issues as well which we vented in a conference call with NetApp. They acknowledged problems and are working to come up with solutions. We'll see. :-) Drew -- "I must obey the inscrutable exhortations of my soul. My mandate also includes weird bugs." -- Calvin

1 0

netapp-top - display current top NFS clients for a NetApp
by Daniel Quinlan 23 Mar '98

23 Mar '98

Here's a useful little script, especially if you ever need to determine which hosts are banging the hardest on an overloaded NetApp. Example output: NFS/s total NFS/s = 2217 | interval = 60 ------ ---------------------------------------- 439 sodium.transmeta.com 437 iridium.transmeta.com 264 tee.transmeta.com 102 manganese.transmeta.com 49 hammer.transmeta.com 43 iron.transmeta.com 41 blighty.transmeta.com 39 curie.transmeta.com ... Before writing this, I usually ended up zeroing the per-client stats or using "diff" between two "nfsstat -l" commands to determine the current top NFS clients. If you make any significant improvements, please send them back to me. Also, please do not post this to Usenet or on a public web site with my email address (spam paranoia). Thanks. - Dan ------- start of cut text -------------- #!/usr/local/bin/perl # netapp-top - display current top NFS clients for a NetApp # Copyright (C) 1998 Daniel Quinlan # # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by # the Free Software Foundation; either version 2 of the License, or # (at your option) any later version. # # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with this program; if not, write to the Free Software # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. $default_interval = 60; $prog = $0; $prog =~ s@.*/@@; require "getopts.pl"; &Getopts("hi:"); if (defined $ENV{"LINES"} && $ENV{"LINES"} ne "") { $lines = $ENV{"LINES"}; } else { $lines = 24; } if ($opt_h) { &usage; exit 0; } if ($opt_i && $opt_i > 0) { $interval = $opt_i; } else { $interval = $default_interval; } if ($ARGV[0]) { $host = $ARGV[0]; } else { &usage; exit 1; } sub usage { print <<EOF; usage: $prog [options] filer -h print this help -i n report filer statistics every n seconds ($default_interval is default) The NetApp must have the "nfs.per_client_stats.enable" option "on" for this to work at all. This program must be run this from an admin host as root. EOF } sub print_top { my $total = 0; my $diff; my %recent; foreach $key (keys %nfs_ops_last) { $diff = $nfs_ops{$key} - $nfs_ops_last{$key}; $recent{$key} = $diff; $total += $diff; } $total = int ($total / $interval); system("clear"); printf "%6s\t%s\n", "NFS/s", "total NFS/s = $total | interval = $interval"; printf "%6s\t%s\n", "------", "----------------------------------------"; $i = 0; foreach $key (sort {$recent{$b} <=> $recent{$a} } keys %recent) { printf "%6d\t%s\n", int ($recent{$key} / $interval + 0.5), $ip_name{$key}; last if ($i++ > ($lines - 5)); } } $first = 1; for (;;) { $time_0 = time(); open(NFSSTAT, "rsh $host nfsstat -l|"); while (<NFSSTAT>) { @tmp = split; # ip_address = tmp[0] # ip_name = tmp[1] # nfs_ops = tmp[4] $ip_name{$tmp[0]} = $tmp[1]; $nfs_ops{$tmp[0]} = $tmp[4]; } close(NFSSTAT); if (! $first) { &print_top; } else { $first = 0; } %nfs_ops_last = %nfs_ops; $time_1 = time(); sleep ($interval - ($time_1 - $time_0)); } ------- end ----------------------------

2 1

inn and 630
by David Power 21 Mar '98

21 Mar '98

Has anybody had any success using a 630 as a back end for inn? We are experiencing some horrible thruput issues with large files. The Inn machine is a multi processor sparc 20s w 512 meg ram running 2.6. Nettapp is a 630 8meg nvram 256meg system ram 100bt full duplex connection between the sparc and the netapp. System load is around 10% Systat 1 shows writes are flushed at around 10 sec intervels, cache age is sitting between 11 and 20. To quote my sys admin. "When INN writes an article, it opens it (O_WRONLY|O_CREAT|O_TRUNC), fills the buffer, then writev's to write out the buffer. This works great on small files, but large files are making the writev just hang for a second or so (not coincidentally at the same time that the dnewslink process sleeps in diablo). I tried changing the open to O_WRONLY|O_NONBLOCK|O_CREAT|O_TRUNC, but apparently non-blocking writes don't happen over NFS. Anyway, the key to solving this is getting the Netapps to write big files faster. Any ideas?" David Power V-P Operations Insync Internet Services Inc.

3 2

None
by Mark D Simmons 20 Mar '98

20 Mar '98

Sorry, didn't cc this properly: --- Forwarded mail from "Mark D Simmons" <mds(a)gbnet.net> >From mds(a)gbnet.net Fri Mar 20 22:02:23 1998 Return-Path: <mds(a)gbnet.net> Delivered-To: mds(a)gbnet.net Received: (qmail 7924 invoked by uid 130); 20 Mar 1998 22:02:23 -0000 Date: 20 Mar 1998 22:02:23 -0000 Message-ID: <19980320220223.7922.qmail(a)ns.gbnet.net> From: "Mark D Simmons" <mds(a)gbnet.net> X-Mailer: Mail User's Shell (7.2.5 10/14/92) To: tonylai <tonylai(a)ms7.mitac.com.tw> Subject: Re: Disk drives prices Cc: mds(a)gbnet.net Because the NetApp complexity and cost is actually superior to all the vendors I mention as my fall-back choices. For many specific tasks, including other file and database servers, we do in fact use Sun and SGI. NetApp boxes provide vast, easily managed pools of disk that I rarely need to intervene with to keep running so their cost- effectiveness wrt my time is very high. Ditto the availability of the data we keep on them. --- End of forwarded message from "Mark D Simmons" <mds(a)gbnet.net>

1 0

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