toasters September 2000

toasters@lists.teaparty.net

122 participants
161 discussions

RE: Deleting snapshot
by Premanshu Jain 09 Sep '00

09 Sep '00

Ndmpd status, doesn't revealany active ndmp sessions. These snapshots date back to 17 Aug. Any other way out..avoiding reboot prem -----Original Message----- From: Steve Kappel [mailto:steve.kappel@raistlin.min.ov.com] Sent: Friday, September 08, 2000 1:22 PM To: mrb(a)netapp.com; PrJain(a)shastanets.com Cc: toasters(a)mathworks.com Subject: RE: Deleting snapshot No, not a chance that it is the backup software (NDMP client). Consider this - if your network link goes down your backup (NDMP client) is aborted. Since you have no network link there is no way the backup product can do anything in NDMP to abort or cleanup the backup running in the filer. The filer has to be able to clean up from the NDMP session socket "going away". Since you have "failed backup processes" the NDMP session is closed on the client side. So it is OnTap that is not functioning correctly. You might try looking for the NDMP sessions with "ndmpd status" on the filer and then kill any that are running with "ndmpd kill <sessionid>". >From: Premanshu Jain <PrJain(a)shastanets.com> >To: "'mrb(a)netapp.com'" <mrb(a)netapp.com>, Premanshu Jain <PrJain(a)shastanets.com> >Cc: toasters(a)mathworks.com >Subject: RE: Deleting snapshot >Date: Fri, 8 Sep 2000 11:09:01 -0700 > >NDMP or Veritas Netbackup. All these snapshots were a part of failed backups >prem > >-----Original Message----- >From: J. Bryan Kelsch [mailto:mrb@netapp.com] >Sent: Friday, September 08, 2000 11:13 AM >To: Premanshu Jain >Cc: toasters(a)mathworks.com >Subject: Re: Deleting snapshot > > >This is not a function the filer, it is an error in the backup software that >is not cleaning up properly. > > > > >> >> >> >>You have to reboot first. >> >> >>On Fri, 8 Sep 2000 10:42:26 -0700 , you wrote: >> >>>I have some snapshots created by failed backup processes and shows them >busy >>>still. How can I delete them..? >>> >>> >>>Volume sw >>>working.................................................................. . >.. >>>......................................................................... . >.. >>>.................................... >>> >>> %/used %/total date name >>>---------- ---------- ------------ -------- >>> 3% ( 3%) 2% ( 2%) Sep 07 18:04 bot_sw.1 >>> 4% ( 3%) 3% ( 2%) Sep 07 16:12 hourly.0 >>> 4% ( 4%) 3% ( 3%) Sep 07 12:10 hourly.1 >>> 5% ( 4%) 3% ( 3%) Sep 07 08:10 hourly.2 >>> 5% ( 4%) 4% ( 3%) Sep 07 00:11 nightly.0 >>> 6% ( 4%) 4% ( 3%) Sep 06 20:11 hourly.3 >>> 7% ( 4%) 5% ( 3%) Sep 06 16:12 hourly.4 >>> 7% ( 4%) 5% ( 3%) Sep 06 12:10 hourly.5 >>> 8% ( 4%) 6% ( 3%) Sep 06 00:08 nightly.1 >>> 19% (15%) 17% (12%) Aug 20 21:52 snapshot_for_backup.32 (busy) >>> 20% (15%) 17% (12%) Aug 19 22:46 snapshot_for_backup.30 (busy) >>> 21% (15%) 19% (12%) Aug 17 04:08 snapshot_for_backup.23 (busy) > > >------------------------ >J. Bryan Kelsch, PSE >Network Appliance >5220 Spring Valley Road - Suite 350 >Dallas, TX 75240 >972.759.7528 Office >817.821.0350 Cell >888.374.6422 Pager __________________________________________________________________________ Steve Kappel steve.kappel(a)veritas.com VERITAS Software steve.kappel(a)iname.com (Personal)

3 2

multiple raid grps in filesystem on filer
by neil lehrer 08 Sep '00

08 Sep '00

hi, i understand that having more than 1 raid grp in a filesystem makes the filer more fault tolerant of a full raid grp failure. so lets say there are two raid grps in the filesystem. if you lose a raid grp [2 drives fail] how do you restore it? i can understand restoring the whole filesystem or a qtree or a dir, but if files and dirs can go across more than 1 raid grp in a filesystem what is the advantage of short restore of the broken raid grp? how do you put back just that raid group? i know you can replace the bad drives and then recreate the group but how do you restore the data for just that raid grp? thanks

2 1

RE: Deleting snapshot
by Steve Kappel 08 Sep '00

08 Sep '00

No, not a chance that it is the backup software (NDMP client). Consider this - if your network link goes down your backup (NDMP client) is aborted. Since you have no network link there is no way the backup product can do anything in NDMP to abort or cleanup the backup running in the filer. The filer has to be able to clean up from the NDMP session socket "going away". Since you have "failed backup processes" the NDMP session is closed on the client side. So it is OnTap that is not functioning correctly. You might try looking for the NDMP sessions with "ndmpd status" on the filer and then kill any that are running with "ndmpd kill <sessionid>". >From: Premanshu Jain <PrJain(a)shastanets.com> >To: "'mrb(a)netapp.com'" <mrb(a)netapp.com>, Premanshu Jain <PrJain(a)shastanets.com> >Cc: toasters(a)mathworks.com >Subject: RE: Deleting snapshot >Date: Fri, 8 Sep 2000 11:09:01 -0700 > >NDMP or Veritas Netbackup. All these snapshots were a part of failed backups >prem > >-----Original Message----- >From: J. Bryan Kelsch [mailto:mrb@netapp.com] >Sent: Friday, September 08, 2000 11:13 AM >To: Premanshu Jain >Cc: toasters(a)mathworks.com >Subject: Re: Deleting snapshot > > >This is not a function the filer, it is an error in the backup software that >is not cleaning up properly. > > > > >> >> >> >>You have to reboot first. >> >> >>On Fri, 8 Sep 2000 10:42:26 -0700 , you wrote: >> >>>I have some snapshots created by failed backup processes and shows them >busy >>>still. How can I delete them..? >>> >>> >>>Volume sw >>>working................................................................... >.. >>>.......................................................................... >.. >>>.................................... >>> >>> %/used %/total date name >>>---------- ---------- ------------ -------- >>> 3% ( 3%) 2% ( 2%) Sep 07 18:04 bot_sw.1 >>> 4% ( 3%) 3% ( 2%) Sep 07 16:12 hourly.0 >>> 4% ( 4%) 3% ( 3%) Sep 07 12:10 hourly.1 >>> 5% ( 4%) 3% ( 3%) Sep 07 08:10 hourly.2 >>> 5% ( 4%) 4% ( 3%) Sep 07 00:11 nightly.0 >>> 6% ( 4%) 4% ( 3%) Sep 06 20:11 hourly.3 >>> 7% ( 4%) 5% ( 3%) Sep 06 16:12 hourly.4 >>> 7% ( 4%) 5% ( 3%) Sep 06 12:10 hourly.5 >>> 8% ( 4%) 6% ( 3%) Sep 06 00:08 nightly.1 >>> 19% (15%) 17% (12%) Aug 20 21:52 snapshot_for_backup.32 (busy) >>> 20% (15%) 17% (12%) Aug 19 22:46 snapshot_for_backup.30 (busy) >>> 21% (15%) 19% (12%) Aug 17 04:08 snapshot_for_backup.23 (busy) > > >------------------------ >J. Bryan Kelsch, PSE >Network Appliance >5220 Spring Valley Road - Suite 350 >Dallas, TX 75240 >972.759.7528 Office >817.821.0350 Cell >888.374.6422 Pager __________________________________________________________________________ Steve Kappel steve.kappel(a)veritas.com VERITAS Software steve.kappel(a)iname.com (Personal)

2 1

RE: Deleting snapshot
by Premanshu Jain 08 Sep '00

08 Sep '00

Oh..This is CRAZY on part of NetAPPS...!!!!!!!!!! -----Original Message----- From: J. Bryan Kelsch [mailto:mrb@netapp.com] Sent: Friday, September 08, 2000 10:56 AM To: Premanshu Jain Subject: Re: Deleting snapshot You have to reboot first. On Fri, 8 Sep 2000 10:42:26 -0700 , you wrote: >I have some snapshots created by failed backup processes and shows them busy >still. How can I delete them..? > > >Volume sw >working.................................................................... . >........................................................................... . >.................................... > > %/used %/total date name >---------- ---------- ------------ -------- > 3% ( 3%) 2% ( 2%) Sep 07 18:04 bot_sw.1 > 4% ( 3%) 3% ( 2%) Sep 07 16:12 hourly.0 > 4% ( 4%) 3% ( 3%) Sep 07 12:10 hourly.1 > 5% ( 4%) 3% ( 3%) Sep 07 08:10 hourly.2 > 5% ( 4%) 4% ( 3%) Sep 07 00:11 nightly.0 > 6% ( 4%) 4% ( 3%) Sep 06 20:11 hourly.3 > 7% ( 4%) 5% ( 3%) Sep 06 16:12 hourly.4 > 7% ( 4%) 5% ( 3%) Sep 06 12:10 hourly.5 > 8% ( 4%) 6% ( 3%) Sep 06 00:08 nightly.1 > 19% (15%) 17% (12%) Aug 20 21:52 snapshot_for_backup.32 (busy) > 20% (15%) 17% (12%) Aug 19 22:46 snapshot_for_backup.30 (busy) > 21% (15%) 19% (12%) Aug 17 04:08 snapshot_for_backup.23 (busy) ------------------------ J. Bryan Kelsch, PSE Network Appliance 5220 Spring Valley Road - Suite 350 Dallas, TX 75240 972.759.7528 Office 817.821.0350 Cell 888.374.6422 Pager

2 1

Deleting snapshot
by Premanshu Jain 08 Sep '00

08 Sep '00

I have some snapshots created by failed backup processes and shows them busy still. How can I delete them..? Volume sw working..................................................................... ............................................................................ .................................... %/used %/total date name ---------- ---------- ------------ -------- 3% ( 3%) 2% ( 2%) Sep 07 18:04 bot_sw.1 4% ( 3%) 3% ( 2%) Sep 07 16:12 hourly.0 4% ( 4%) 3% ( 3%) Sep 07 12:10 hourly.1 5% ( 4%) 3% ( 3%) Sep 07 08:10 hourly.2 5% ( 4%) 4% ( 3%) Sep 07 00:11 nightly.0 6% ( 4%) 4% ( 3%) Sep 06 20:11 hourly.3 7% ( 4%) 5% ( 3%) Sep 06 16:12 hourly.4 7% ( 4%) 5% ( 3%) Sep 06 12:10 hourly.5 8% ( 4%) 6% ( 3%) Sep 06 00:08 nightly.1 19% (15%) 17% (12%) Aug 20 21:52 snapshot_for_backup.32 (busy) 20% (15%) 17% (12%) Aug 19 22:46 snapshot_for_backup.30 (busy) 21% (15%) 19% (12%) Aug 17 04:08 snapshot_for_backup.23 (busy)

2 1

RE: Deleting snapshot
by Premanshu Jain 08 Sep '00

08 Sep '00

NDMP or Veritas Netbackup. All these snapshots were a part of failed backups prem -----Original Message----- From: J. Bryan Kelsch [mailto:mrb@netapp.com] Sent: Friday, September 08, 2000 11:13 AM To: Premanshu Jain Cc: toasters(a)mathworks.com Subject: Re: Deleting snapshot This is not a function the filer, it is an error in the backup software that is not cleaning up properly. > > > >You have to reboot first. > > >On Fri, 8 Sep 2000 10:42:26 -0700 , you wrote: > >>I have some snapshots created by failed backup processes and shows them busy >>still. How can I delete them..? >> >> >>Volume sw >>working................................................................... .. >>.......................................................................... .. >>.................................... >> >> %/used %/total date name >>---------- ---------- ------------ -------- >> 3% ( 3%) 2% ( 2%) Sep 07 18:04 bot_sw.1 >> 4% ( 3%) 3% ( 2%) Sep 07 16:12 hourly.0 >> 4% ( 4%) 3% ( 3%) Sep 07 12:10 hourly.1 >> 5% ( 4%) 3% ( 3%) Sep 07 08:10 hourly.2 >> 5% ( 4%) 4% ( 3%) Sep 07 00:11 nightly.0 >> 6% ( 4%) 4% ( 3%) Sep 06 20:11 hourly.3 >> 7% ( 4%) 5% ( 3%) Sep 06 16:12 hourly.4 >> 7% ( 4%) 5% ( 3%) Sep 06 12:10 hourly.5 >> 8% ( 4%) 6% ( 3%) Sep 06 00:08 nightly.1 >> 19% (15%) 17% (12%) Aug 20 21:52 snapshot_for_backup.32 (busy) >> 20% (15%) 17% (12%) Aug 19 22:46 snapshot_for_backup.30 (busy) >> 21% (15%) 19% (12%) Aug 17 04:08 snapshot_for_backup.23 (busy) ------------------------ J. Bryan Kelsch, PSE Network Appliance 5220 Spring Valley Road - Suite 350 Dallas, TX 75240 972.759.7528 Office 817.821.0350 Cell 888.374.6422 Pager

1 0

RE: vif between 740 & 520 problems & questions
by Thompson, Arnold 08 Sep '00

08 Sep '00

Hi John, You're not doing anything wrong - it's the load-balancing algorithm used by etherchannel that's causing you to use only one interface. In a nutshell, etherchannel does a hash on the last two bits of the source MAC address and comes up with one of four possible combinations. Your NDMPcopy connection starts up on only one port, using a single MAC address, so essentially there are no multiple MAC addresses to load-balance. Additionally, the balancing decision based on this hash is made by an etherchannel-enabled switch, not the Netapp box itself. Etherchannel requires all ports to be on the same switch and only really works when you have multiple clients (and consequently multiple possible hash combinations) accessing one server - it fails completely when you have a single "client" accessing a single server. In brief, to get the best migration performance, use gigabit ethernet between two systems. You can simply reverse the optical cable leads - you don't even need a switch. The good news is that more sophisticated load-balancing algorithms are under consideration, although I can't give you specifics or tell you when Netapp will have them available. Good luck, Arnie > -----Original Message----- > From: John Stoffel [mailto:john@casc.com] > Sent: Friday, September 08, 2000 7:11 AM > To: toasters(a)mathworks.com > Subject: vif between 740 & 520 problems & questions > > > > Hi all, > > I've got a couple of questions on setting up and using vifs between > two filers to help speed up data migration from the old F520 to a new > F740. > > They're configed like this: > > F520 F740 > OnTap: 5.2.1 OnTap: 5.3.6p2 > RAM: 256Mb RAM: 512Mb > Quad Ethernet Quad Ethernet > Gigabit Ethernet > > I've networked the two systems together using three ports on each Quad > Ethernet board, but due to distance limitations, I'm going through > three seperate hubs. So the two filers are basically hooked up > back-to-back over a private network. > > I did the various vif commands, and they are setup correctly, and they > are passing traffic properly, but just over one of the three links. I > know I need to upgrade the F520 to match OnTap versions, and will do > so next weekend when I have some down time, but I wanted to play > around with ndmpcopy beforehand to see how the performance will be. > > Doing a test this morning, I managed to crash the new F740 with a PCI > Parity error while doing another test of ndmpcopy. Whee! > > My questions: > > - will vifs work properly across direct connections between filers? > > - Has anyone used a Lucent Cajun P220 switch to implement a VIF? I > could easily run a fiber link from the F740 to a Cajun which would > then run the three ethernet links. > > And while I work for Lucent, the Cajun wasn't made in our division and > I haven't used it. > > Thanks, > John > John Stoffel - Senior Unix Systems Administrator - Lucent > Technologies > stoffel(a)lucent.com - http://www.lucent.com - 978-952-7548 > john.stoffel(a)ascend.com - http://www.ascend.com >

1 0

SUMMARY: vif between 740 & 520 problems & questions
by John Stoffel 08 Sep '00

08 Sep '00

Hi all, The general consensus is that a VIF won't help me at all, since all the data will just go via the primary interface, and won't use the two other interfaces at all. This is what I'm seeing on the system anyway, so I guess I'm out of luck. Thanks for all your comments! John John Stoffel - Senior Unix Systems Administrator - Lucent Technologies stoffel(a)lucent.com - http://www.lucent.com - 978-952-7548 john.stoffel(a)ascend.com - http://www.ascend.com

1 0

vif between 740 & 520 problems & questions
by John Stoffel 08 Sep '00

08 Sep '00

Hi all, I've got a couple of questions on setting up and using vifs between two filers to help speed up data migration from the old F520 to a new F740. They're configed like this: F520 F740 OnTap: 5.2.1 OnTap: 5.3.6p2 RAM: 256Mb RAM: 512Mb Quad Ethernet Quad Ethernet Gigabit Ethernet I've networked the two systems together using three ports on each Quad Ethernet board, but due to distance limitations, I'm going through three seperate hubs. So the two filers are basically hooked up back-to-back over a private network. I did the various vif commands, and they are setup correctly, and they are passing traffic properly, but just over one of the three links. I know I need to upgrade the F520 to match OnTap versions, and will do so next weekend when I have some down time, but I wanted to play around with ndmpcopy beforehand to see how the performance will be. Doing a test this morning, I managed to crash the new F740 with a PCI Parity error while doing another test of ndmpcopy. Whee! My questions: - will vifs work properly across direct connections between filers? - Has anyone used a Lucent Cajun P220 switch to implement a VIF? I could easily run a fiber link from the F740 to a Cajun which would then run the three ethernet links. And while I work for Lucent, the Cajun wasn't made in our division and I haven't used it. Thanks, John John Stoffel - Senior Unix Systems Administrator - Lucent Technologies stoffel(a)lucent.com - http://www.lucent.com - 978-952-7548 john.stoffel(a)ascend.com - http://www.ascend.com

1 0

FC card slot exchange helped us [Was : gigabit ethernet performance]
by kubek 08 Sep '00

08 Sep '00

I do not have a solution but maybe these experiments done with a 720 may help (I must thank our local reseller support here :). We received an F720 some weeks ago (quad ethernet / FC9) A 'systat 1' showed a really bad performance on the administrative Ethernet (100M) interface : the filer was inputting ~4MB/s when creating a 100MB file from a Sparc. By moving the FC card from the slot 8 to the slot 2 and rebooting the (little:() beast we are now going to ~11MB/s . Seems that netapp hardware layer is as poor as some PC cards builders :( Maybe the trick will work for you : try to move your FC card from its current slot (8?) to another one. Jean-Marie

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 September 2000