toasters May 2002

toasters@lists.teaparty.net

62 participants
83 discussions

mixed security style
by Kelly McQuarrie 07 May '02

07 May '02

Hello all: I have a large number of windows/unix home directories on my NetApps all under unix qtrees. I'm a unix admin so this works great for me but the NT admins have complained that they can access their users data. I would like any war stories, opions, etc from people who have tried and or use the mixed qtrees for this purpose. Any suggestions and what not are totally welcomed! Thank you sooo much for your time 8) Kelly McQuarrie Unix System Administrator Ericsson CDMA Systems

3 2

RE: FW: mixed security style
by Lupa Paul-ra5507 07 May '02

07 May '02

Mixed Security provides the fine grain access control that I need for various security issues. However if a tree has NT security (with inheritance) a number of UNIX problems rear their ugly head, with UNIX security, I don't have the control that I need. I guess the only solution is to wait for MicroSoft to take over the world, or Sun/HP/Linux to support NFS version 4 (I'm not holding my breath). My current issue centers around gzip/compress which will not work if inheritance is turned on. The different paradigm for security between NT-acls and standard UNIX makes for an ugly solution. Does EMC have a solution?... Cheers, Paul Lupa Sr.System Analyst Motorola, Austin -----Original Message----- From: Kelly McQuarrie [mailto:kellym@sd.us.am.ericsson.se] Sent: Tuesday, May 07, 2002 3:29 PM To: Maglinger, Paul Cc: 'NetApps list server' Subject: Re: FW: mixed security style People running stuff from unix on an NTFS file system don't run into problems? Do they every lose access to there files from Unix because of changes made from NT? Kelly McQuarrie Unix System Administrator Ericsson CDMA Systems On Tue, 7 May 2002, Maglinger, Paul wrote: > We have mixed NT and Unix environment. If the qtree is strictly Unix, we > set the style for Unix. If both use it, then I have it set for NTFS. We > tried running mixed and sometimes Unix could access it and sometimes it > couldn't. Seemed to be the same with NT, just kinda flaky stuff. I'm an NT > admin, so this works great for me <grin>, but the Unix admin doesn't seem to > have a problem with it set up this way... We also disabled the oplocks. > > -----Original Message----- > From: Kelly McQuarrie [mailto:kellym@sd.us.am.ericsson.se] > Sent: Tuesday, May 07, 2002 14:06 > To: toasters(a)mathworks.com > Subject: mixed security style > > > Hello all: > > I have a large number of windows/unix home directories on my NetApps all > under unix qtrees. I'm a unix admin so this works great for me but the NT > admins have complained that they can access their users data. I would > like any war stories, opions, etc from people who have tried and or use > the mixed qtrees for this purpose. Any suggestions and what not are > totally welcomed! > > Thank you sooo much for your time 8) > > Kelly McQuarrie > Unix System Administrator > Ericsson CDMA Systems >

1 0

FW: mixed security style
by Maglinger, Paul 07 May '02

07 May '02

We have mixed NT and Unix environment. If the qtree is strictly Unix, we set the style for Unix. If both use it, then I have it set for NTFS. We tried running mixed and sometimes Unix could access it and sometimes it couldn't. Seemed to be the same with NT, just kinda flaky stuff. I'm an NT admin, so this works great for me <grin>, but the Unix admin doesn't seem to have a problem with it set up this way... We also disabled the oplocks. -----Original Message----- From: Kelly McQuarrie [mailto:kellym@sd.us.am.ericsson.se] Sent: Tuesday, May 07, 2002 14:06 To: toasters(a)mathworks.com Subject: mixed security style Hello all: I have a large number of windows/unix home directories on my NetApps all under unix qtrees. I'm a unix admin so this works great for me but the NT admins have complained that they can access their users data. I would like any war stories, opions, etc from people who have tried and or use the mixed qtrees for this purpose. Any suggestions and what not are totally welcomed! Thank you sooo much for your time 8) Kelly McQuarrie Unix System Administrator Ericsson CDMA Systems

2 1

Error on Gigabit Card
by Manoj Kolekar 06 May '02

06 May '02

Hello toasters, I am getting the bellow Erros onn my F840 gigabit card, "Mon May 6 10:46:46 GMT [GbE-II/e4:error]: arp: 00:06:5b:3a:1d:0c attempts to modify permanent entry for 192.168.57.63 on e4 Mon May 6 10:51:29 GMT [GbE-II/e4:error]: arp: 00:06:5b:3a:1d:0c attempts to modify permanent entry for 192.168.57.63 on e4 Mon May 6 10:51:38 GMT [GbE-II/e4:error]: arp: 00:06:5b:3a:1d:0c attempts to modify permanent entry for 192.168.57.63 on e4 The filer seems to be working fine. Does any one out there have any idea as to what could be wrong. -- Best regards, Manoj mailto:manoj@apara.com

2 1

RE: NFS Stats Questions
by Eisler, Alex 06 May '02

06 May '02

Try setting the link options on the filer to 100/full. Cisco switches are very vulnerable in that department. On the switch (when in enabled mode): 'set port mod/port speed 100' 'set port mod/port duplex full' On the filer: 'ifconfig interface mediatype 100tx-fd' Don't forget to change the ifconfig line in the rc file also... I've seen it happen more than once that synch-negotiation problems deteriorate performance very noticeably... Alex. -----Original Message----- From: Chuah, Nicholas [mailto:nicholas_chuah@yahoo.com] Sent: Monday, May 06, 2002 14:16 To: toasters(a)mathworks.com Subject: NFS Stats Questions Hi all, Unit 720 with on-board 100BaseT, DataONTAP 6.1, NFS over UDP 32K, all Sun clients. Performance degradation was obvious after the switch was upgraded to the Cisco 6050. Had been testing with the TCP changes and changing the UDP datagram size to 8K but no significant improvement. Is there any tweaking I can try before I attempt to add a quad card or gigabit card? What can I deduce from the ifstat below with regards to the list overflow, underflow value? I'm not sure whether it is an acceptable amount of errors or an obvious performance issue given the amount of data transferred. Another query I have is the "high" misses of the NFS cache reply stats. I would appreciate any input to the above queries. Regards, Nic ********* RECEIVE Packets: 3634m | Bytes: 1421g | Total errors: 0 No buffers: 0 | Length errors: 0 | Runt frames: 0 Long frames: 0 | Collisions: 0 | CRC errors: 0 H/w overflow: 136 | List overflow: 44048 | Rst frame drops: 0 TRANSMIT Packets: 6606m | Bytes: 6761g | Total errors: 37 Collisions: 0 | Late collisions: 0 | Excess coll: 0 Queue overflow: 3913 | No carrier: 0 | H/w underflow: 11 Defer: 0 | Link failure: 25 | Time out: 0 List underflow: 1515 | Loss of carrier: 0 | No buffers: 0 DEVICE Resets: 2 LINK INFO Auto: Disabled | Mediatype: 100tx-fd | Link Partner: Unknown | Link State: Up NFS reply cache statistics: TCP: In progress Delay hits Misses Idempotent Non-idempotent 0 0 6075661 0 0 UDP: In progress Delay hits Misses Idempotent Non-idempotent 3233 6072 642460762 5550 1024

2 1

PDC
by Peter Valian 06 May '02

06 May '02

Is anyone using something other than windows to be a PDC? i.e. LSI TotalNet, Sun PC Netlink, Samba? Any suggestions appreciated. -peter -- Peter Valian Network & Systems Administrator Southwestern University Georgetown, Texas --

2 1

NFS Stats Questions
by Chuah, Nicholas 06 May '02

06 May '02

Hi all, Unit 720 with on-board 100BaseT, DataONTAP 6.1, NFS over UDP 32K, all Sun clients. Performance degradation was obvious after the switch was upgraded to the Cisco 6050. Had been testing with the TCP changes and changing the UDP datagram size to 8K but no significant improvement. Is there any tweaking I can try before I attempt to add a quad card or gigabit card? What can I deduce from the ifstat below with regards to the list overflow, underflow value? I'm not sure whether it is an acceptable amount of errors or an obvious performance issue given the amount of data transferred. Another query I have is the "high" misses of the NFS cache reply stats. I would appreciate any input to the above queries. Regards, Nic ********* RECEIVE Packets: 3634m | Bytes: 1421g | Total errors: 0 No buffers: 0 | Length errors: 0 | Runt frames: 0 Long frames: 0 | Collisions: 0 | CRC errors: 0 H/w overflow: 136 | List overflow: 44048 | Rst frame drops: 0 TRANSMIT Packets: 6606m | Bytes: 6761g | Total errors: 37 Collisions: 0 | Late collisions: 0 | Excess coll: 0 Queue overflow: 3913 | No carrier: 0 | H/w underflow: 11 Defer: 0 | Link failure: 25 | Time out: 0 List underflow: 1515 | Loss of carrier: 0 | No buffers: 0 DEVICE Resets: 2 LINK INFO Auto: Disabled | Mediatype: 100tx-fd | Link Partner: Unknown | Link State: Up NFS reply cache statistics: TCP: In progress Delay hits Misses Idempotent Non-idempotent 0 0 6075661 0 0 UDP: In progress Delay hits Misses Idempotent Non-idempotent 3233 6072 642460762 5550 1024

1 0

two nics on same subnet
by Piotr KUCHARSKI 04 May '02

04 May '02

Having atm and fe I wanted the latter to serve as a backup in case atm fails. So I configured atm to be .7 and fe to be .6 in the same subnet. It gave me a warning: ifconfig (warning): route already exists (probably due to another interface on the same subnet.) but that I (almost) expected. However, when I test it and pull out the atm, fe doesn't respond to me (quite successfully destroying my concept of backup/load balancing/automounter). That's not what I expected. Any thoughts? p.

2 2

long cifs.home_dir
by Piotr KUCHARSKI 03 May '02

03 May '02

I have 14 qtrees and wanted all of them to be in cifs.home_dir option, started from /vol/vol0/stud1994/%u/NT... and it said error: option value length too long so i renamed qtrees to shorter names, like s94, renamed vol0 to v and now it fits. But what if I have more qtrees? Any way to enlarge space for it? (I thought of /vol/vol0/*/%u%/NT, but didn't work) p.

2 2

OnTap SNMP Traps
by Ben Rockwood 03 May '02

03 May '02

Hello Toasterites, I've been (trying) to experiment with SNMP traps with a test filer I have. However, I've run into a road block, or so I think. Is it not true that OnTap should have a series of predefined traps? In the documentation it suggests using default traps by loading "/etc/defaults/traps", however a /etc/defaults directory doesn't exsist on any of the filers I have. Does this actually mean that if I want to add traps I should be placing them in /etc/default/traps instead of another file? I find it odd that OnTap would make me custom define _all_ of my traps (minus the standard traps defined in RFC1213). I'll even note that on page 563 of the OnTap 5.1 users guide it mentions there being Builtin traps as defined in the MIB for diskFailedShutdown,etc. So far, the only traps I've been sent from my test filer is "coldStarts" after I initialize SNMP traps (snmp init 1). Any ideas? Ben Rockwood

1 0

← Newer
1
2
3
4
5
6
7
8
9
Older →

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 May 2002