toasters

toasters@lists.teaparty.net

2 participants
13520 discussions

Solaris 2.6 drops NFS automounts
by Mike Mueller 23 Jan '98

23 Jan '98

We have heard reports that Solaris 2.6 NFS clients drop mounts of NetApp and other NFS servers but I have not been able to get a detailed explanation of the problem. Although it is apparently a Solaris 2.6 bug, I can't seem to find information about it from Sun and I thought that some other NetApp user might be familiar with the problem. We are planning a Solaris upgrade in the next few weeks and I'd like to feel confident that we aren't going to run into problems. Can someone explain what the specific bug is and whether there is a definate work-around or patch available? Is anyone successfully using Solaris 2.6 clients with a NetApp and if so are you using any special settings on either the client and/or the server? Thanks. Mike Mueller

2 1

Re: Anybody besides me having quotas go bonkers?
by Jan-Pieter Cornet 23 Jan '98

23 Jan '98

In article <199801202125.NAA26037(a)tooting.netapp.com>, Rajesh Sundaram <rajeshs(a)netapp.com> wrote: > attribution missing, but someone(a)cdsnet.net wrote: >> I just created some new users, and the filer (running 4.3R4D1) >> is reporting that they've used: >> >> user 506 9840 41943040 1292 10485760 >> user 1001 4294966368 41943040 22 10485760 >> user@web 1001 4294966408 25600 18 1000 >> user 1003 4294967260 41943040 1 10485760 >> user@web 1003 4294967260 25600 0 1000 >> user 1004 4294967260 41943040 1 10485760 >> user@web 1004 4294967260 25600 0 1000 >> >> which is flatly impossible, the users were just created and assigned. yet >> it thinks the disk space usage is all used up. We have been seeing the same thing last week. We fixed it by putting a (root-owner) file in the users' homedirectory, then chowning that file to the user, making their disk usage positive again. >Are default quotas being used here? If that's the case, this is likely to >be the bug with operations running as root that others in the list >have pointed out. > >Default quotas trigger the creation of quotas when a user first >uses an inode or a data block (by creating a file, for example). >This should happen even when the file is created as root and then >"chown"ed to the user. This currently fails to happen. > >Here is the description from the bug report which should soon get >added to the bugs online list on NOW. > >TITLE: >Derived quotas do not get created for operations >running as root, causing usage decrepancies. We upgraded to 4.3R4D2, which apparently fixed it. At least, I haven't seen "negative" disk usages in almost a day, whereas our previous version (4.3R4D1) would have had some by now. >> What's even weirder is that I can create directories all day, and no >> errors, only files cause problems. yet the dirs are owned by user 1001, >> etc. Toasters do not count directory blocks as disk usage, so directories can't be affected by this bug. -- #! ##### Jan-Pieter Cornet ##### <johnpc(a)xs4all.net> ##### perl ++$_;$!=$_+++$_;($:,$,,$/,$*)=$!=~/.(.)...(.)(.).(.)/;$!=$_+$_; ($(a),$\,$~)=$!=~/(.)(.).(.)/; $_="$,$/$:"; $@++; $~="$~$_";($_)= \$$=~/\((.)/;$|=++$_;$_++;$|++;$~="$~ $@$:";`$~$/$\$*$, $|>&$_`

1 0

4.3R4 and statd
by Michael Salmon 22 Jan '98

22 Jan '98

We have just installed 4.3R4 in a 520 with storage shelves 2 and statd has begun complaining about the shelves. Shelfchk has also stopped working so we are forced to conclude that the connection between the filer and the shelves has stopped working. Any comments or suggestions? /Michael -- /--------------------------------\ ___ | Michael Salmon | / \ | Ericsson Utveckling AB | / | | ÄS/UAB/Z/IPE | / _| | Box 1505 | / | | 125 25 Älvsjö | / / | Sweden | / / | | | | +--------------------------------+ | \ | Michael.Salmon(a)uab.ericsson.se | / *| | | / _/ | Tel: +46 8 727 30 74 | | | | FAX: +46 8 647 82 76 | | / # | GSM: +46 705 902 842 | | __// \--------------------------------/ |_/

1 0

Archived mail
by Dawn Hamel 22 Jan '98

22 Jan '98

Is the mail from this list archived anywhere. I believe there was some discussion about problems upgrading to Solaris 2.6.1 and I want to make sure a new Netapp customer is aware of any problems BEFORE they upgrade to the 2.6.1 release Thanks, Dawn

8 9

Anybody besides me having quotas go bonkers?
by Jaye Mathisen 20 Jan '98

20 Jan '98

I have a few quota trees set up on a 540 with 100GB. I just created some new users, and the filer (running 4.3R4D1) is reporting that they've used: user 506 9840 41943040 1292 10485760 user 1001 4294966368 41943040 22 10485760 user@web 1001 4294966408 25600 18 1000 user 1003 4294967260 41943040 1 10485760 user@web 1003 4294967260 25600 0 1000 user 1004 4294967260 41943040 1 10485760 user@web 1004 4294967260 25600 0 1000 which is flatly impossible, the users were just created and assigned. yet it thinks the disk space usage is all used up. What's even weirder is that I can create directories all day, and no errors, only files cause problems. yet the dirs are owned by user 1001, etc. If anybody has any ideas, I'd love to hear it, occasionally working quotas is worse than no quotas, since it prevents mail from being deliverd (using qmail), yet seems to run OK. My quota file is trivial, just a few entries. CC'd to support@netapp, because I have a separate open ticket where df will show -8kb used, -8kb free at odd times. Any tips appreciated, I could almost strangle this thing.

2 1

Re: Anybody besides me having quotas go bonkers?
by Daniel Quinlan 19 Jan '98

19 Jan '98

Jaye Mathisen <mrcpu(a)cdsnet.net> writes: > which is flatly impossible, the users were just created and assigned. yet > it thinks the disk space usage is all used up. We had a similar problem with some 4.2 versions. Basically, files would get created as root and then chowned to a user, but without any quota usage attribution. Later, the user deletes some files. 0 minus some number is a really big number because of integer overflow. Quota exceeded. It looks like you may have a similar problem. Generally, quotas are unreliable for us. I always "quota off; quota on" whenever I add a new directory directly under one of our quota trees. The biggest problem for us is that we don't really want to use quotas to limit disk usage, we just want to track it. It should be really easy to have a "nice quotas" option that causes the filer to never send an EDQUOT (quota exceeded) error. That one feature would make most of our quota problems non-problems. - Dan

2 2

None
by Kelly Wyatt 16 Jan '98

16 Jan '98

subscribe toasters

1 0

Shortcomings of current NetApps?
by Marc Nicholas 15 Jan '98

15 Jan '98

I'm interested to hear what others think the shortcomings of the current crop of NetApps are. From my experience: * Single logical drive * Weedy SCSI performance * Lack of other RAID implementations * Not quite enough redundancy for truly mission critical applications Don't get me wrong, I love 'em :-) -marc --- Marc Nicholas - Hippocampus OSD, Inc. - Eastern Office 416 979 9000 - fax: 416 979 8223 - http://www.hippocampus.net 125 John St. - Suite #100 - Toronto - Ontario - M5V 2E2 - CANADA "Inter/Intra/Extra[net] consulting, corporate access, hardware and software sales"

14 27

Re: Shortcomings of current NetApps?
by Daniel Quinlan 15 Jan '98

15 Jan '98

rasmus(a)lerdorf.on.ca (Rasmus Lerdorf) writes: > > > 1. Out-of-date SNMP implemention - (no CIFS or disk system statistics, > > for example) > > What do you mean by disk system stats here? The information displayed by the "sysstat" command. You can get the NFS and CPU load via SNMP, but that's it. CPU NFS CIFS HTTP Net kB/s Disk kB/s Tape kB/s Cache in out read write read write age --- --- --- ---- Only the items with dashes under them are available via SNMP. Dan

1 0

Re: Shortcomings of current NetApps?
by mds46523＠ggr.co.uk 15 Jan '98

15 Jan '98

On Jan 15, 8:46, Brian Tao wrote: > Subject: Re: Shortcomings of current NetApps? > I have a really simple request: a local "rm" command that can > take arguments either on the command line or from stdin. Anyone > running a traditional news spool on a Netapp will probably appreciate > what I'm getting at. ;-) A local "du" command would be rather spiffy too. A special du keyed to UID, ie the space used by a person in a given directory tree would also be kinda cool. I think that there are certain operations where having the filer do the whole thing would save the network, your client machine and the filer itself a lot of time and probably get the job done quicker. The operation could probably be made as low priority as you like and it would still have advantages. Maybe if there were a JAVA VM on the filer that could manipulate the filesystem consistently we could have a bash ourselves. Clearly this could be open to abuse of the Appliance principle, but that's where discipline comes in. > -- > Brian Tao (BT300, taob(a)netcom.ca) > "Though this be madness, yet there is method in't" >-- End of excerpt from Brian Tao -- -Mark, TSG Unix admin and support, int 782 4412, ext +44 1438 76 4412. ... an Englishman in London ...

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