Hi all, I''ve got an interesting (I think) thing happening with my storage pool (tank, 8x1.5TB RAID-Z2)...namely that I seem to gain free-space without deleting files. I noticed this happening awhile ago, so I set up a cron script that ran every night and does: pfexec ls -alR /tank > /export/home/movax/fslog/`date +%F`.contents pfexec /usr/sbin/zfs list > /export/home/movax/fslog/`date +%F`.zfs so I could see if files are disappearing. Well, between 10/14 and 10/15, I went from: tank 8.02T 1.64G 44.1M /tank to: tank 8.01T 14.1G 44.1M /tank The only file changes that showed up in a diff are two file modification date display changes (Apr 17 23:13 to Apr 17 2010, whatever no big deal), both to Thumbs.db. So, there doesn''t *seem* to be any data loss (this is my personal fileserver, all other users have read-only access)...and I don''t recall a ZFS garbage-collect or similar mechanism. Here are the options set on tank: NAME PROPERTY VALUE SOURCE tank type filesystem - tank creation Tue Feb 24 4:20 2009 - tank used 8.01T - tank available 14.1G - tank referenced 44.1M - tank compressratio 1.00x - tank mounted yes - tank quota none default tank reservation none default tank recordsize 128K default tank mountpoint /tank default tank sharenfs off default tank checksum on default tank compression off default tank atime on default tank devices on default tank exec on default tank setuid on default tank readonly off default tank zoned off default tank snapdir hidden local tank aclmode groupmask default tank aclinherit restricted local tank canmount on default tank shareiscsi off default tank xattr on default tank copies 1 default tank version 4 - tank utf8only off - tank normalization none - tank casesensitivity sensitive - tank vscan off default tank nbmand off default tank sharesmb off local tank refquota none default tank refreservation none default tank primarycache all default tank secondarycache all default tank usedbysnapshots 87.5K - tank usedbydataset 44.1M - tank usedbychildren 8.01T - tank usedbyrefreservation 0 - tank logbias latency default tank dedup off local tank mlslabel none default tank com.sun:auto-snapshot true local I don''t utilize snapshots (this machine just stores media)...so what could be up? -- --khd
>tank com.sun:auto-snapshot true local > >I don''t utilize snapshots (this machine just stores media)...so what >could be up?You''ve also disabled the time-slider functionality? (automatic snapshots) Casper
Huh, I don''t actually ever recall enabling that. Perhaps that is
connected to the message I started getting every minute recently in
the kernel buffer,
Oct 20 12:20:49 megatron pcplusmp: [ID 805372 kern.info] pcplusmp: ide
(ata) instance 3 irq 0xf vector 0x45 ioapic 0x2 intin 0xf is bound to
cpu 0
Oct 20 12:21:49 megatron pcplusmp: [ID 805372 kern.info] pcplusmp: ide
(ata) instance 3 irq 0xf vector 0x45 ioapic 0x2 intin 0xf is bound to
cpu 1
I just disabled it ("zfs set com.sun\:auto-snapshot=false tank",
correct?), will see if the log messages disappear. Did the filesystem
kill off some snapshots or something in an effort to free up space?
On Wed, Oct 20, 2010 at 12:26 PM, <Casper.Dik at sun.com>
wrote:>
>
>>tank ?com.sun:auto-snapshot ?true ? ? ? ? ? ? ? ? ? local
>>
>>I don''t utilize snapshots (this machine just stores media)...so
what
>>could be up?
>
>
> You''ve also disabled the time-slider functionality? ?(automatic
snapshots)
>
> Casper
>
>
--
--khd
On 20 October, 2010 - Krunal Desai sent me these 1,5K bytes:> Huh, I don''t actually ever recall enabling that. Perhaps that is > connected to the message I started getting every minute recently in > the kernel buffer, > > Oct 20 12:20:49 megatron pcplusmp: [ID 805372 kern.info] pcplusmp: ide > (ata) instance 3 irq 0xf vector 0x45 ioapic 0x2 intin 0xf is bound to > cpu 0 > Oct 20 12:21:49 megatron pcplusmp: [ID 805372 kern.info] pcplusmp: ide > (ata) instance 3 irq 0xf vector 0x45 ioapic 0x2 intin 0xf is bound to > cpu 1 > > I just disabled it ("zfs set com.sun\:auto-snapshot=false tank", > correct?), will see if the log messages disappear. Did the filesystem > kill off some snapshots or something in an effort to free up space?Probably. zfs list -t all to see all the snapshots as well. /Tomas -- Tomas ?gren, stric at acc.umu.se, http://www.acc.umu.se/~stric/ |- Student at Computing Science, University of Ume? `- Sysadmin at {cs,acc}.umu.se
Argh, yes, lots of snapshots sitting around...apparently time-slider got activated somehow awhile back. Disabled the services and am now cleaning out the snapshots! On Wed, Oct 20, 2010 at 12:41 PM, Tomas ?gren <stric at acc.umu.se> wrote:> On 20 October, 2010 - Krunal Desai sent me these 1,5K bytes: > >> Huh, I don''t actually ever recall enabling that. Perhaps that is >> connected to the message I started getting every minute recently in >> the kernel buffer, >> >> Oct 20 12:20:49 megatron pcplusmp: [ID 805372 kern.info] pcplusmp: ide >> (ata) instance 3 irq 0xf vector 0x45 ioapic 0x2 intin 0xf is bound to >> cpu 0 >> Oct 20 12:21:49 megatron pcplusmp: [ID 805372 kern.info] pcplusmp: ide >> (ata) instance 3 irq 0xf vector 0x45 ioapic 0x2 intin 0xf is bound to >> cpu 1 >> >> I just disabled it ("zfs set com.sun\:auto-snapshot=false tank", >> correct?), will see if the log messages disappear. Did the filesystem >> kill off some snapshots or something in an effort to free up space? > > Probably. > > zfs list -t all ? to see all the snapshots as well. > > /Tomas > -- > Tomas ?gren, stric at acc.umu.se, http://www.acc.umu.se/~stric/ > |- Student at Computing Science, University of Ume? > `- Sysadmin at {cs,acc}.umu.se >-- --khd
Krunal, The file system size changes are probably caused when these snapshots are created and deleted automatically. The recurring messages below are driver related and probably have nothing to do with the snapshots. Thanks, Cindy On 10/20/10 10:50, Krunal Desai wrote:> Argh, yes, lots of snapshots sitting around...apparently time-slider > got activated somehow awhile back. Disabled the services and am now > cleaning out the snapshots! > > On Wed, Oct 20, 2010 at 12:41 PM, Tomas ?gren <stric at acc.umu.se> wrote: >> On 20 October, 2010 - Krunal Desai sent me these 1,5K bytes: >> >>> Huh, I don''t actually ever recall enabling that. Perhaps that is >>> connected to the message I started getting every minute recently in >>> the kernel buffer, >>> >>> Oct 20 12:20:49 megatron pcplusmp: [ID 805372 kern.info] pcplusmp: ide >>> (ata) instance 3 irq 0xf vector 0x45 ioapic 0x2 intin 0xf is bound to >>> cpu 0 >>> Oct 20 12:21:49 megatron pcplusmp: [ID 805372 kern.info] pcplusmp: ide >>> (ata) instance 3 irq 0xf vector 0x45 ioapic 0x2 intin 0xf is bound to >>> cpu 1 >>> >>> I just disabled it ("zfs set com.sun\:auto-snapshot=false tank", >>> correct?), will see if the log messages disappear. Did the filesystem >>> kill off some snapshots or something in an effort to free up space? >> Probably. >> >> zfs list -t all to see all the snapshots as well. >> >> /Tomas >> -- >> Tomas ?gren, stric at acc.umu.se, http://www.acc.umu.se/~stric/ >> |- Student at Computing Science, University of Ume? >> `- Sysadmin at {cs,acc}.umu.se >> > > >
>Huh, I don''t actually ever recall enabling that. Perhaps that is >connected to the message I started getting every minute recently in >the kernel buffer,It''s on by default. You can see if it was ever enabled by using: zfs list -t snapshot |grep @zfs-auto>Oct 20 12:20:49 megatron pcplusmp: [ID 805372 kern.info] pcplusmp: id>e >(ata) instance 3 irq 0xf vector 0x45 ioapic 0x2 intin 0xf is bound to >cpu 0 >Oct 20 12:21:49 megatron pcplusmp: [ID 805372 kern.info] pcplusmp: id>e >(ata) instance 3 irq 0xf vector 0x45 ioapic 0x2 intin 0xf is bound to >cpu 1This sounds more like a device driver unloaded and later it is reloaded because of some other service.>I just disabled it ("zfs set com.sun\:auto-snapshot=3Dfalse tank", >correct?), will see if the log messages disappear. Did the filesystem >kill off some snapshots or something in an effort to free up space?Yes, but typically it will log that. Casper
Where would that log be located? Tried poking around in /var/svc/log and /var/adm, but I''ve found just the snapshot-service logs (while useful, they don''t seem to have logged the auto-deletion of snapshots). Also, that ''pcplusmp'' is triggering every minute, on the minute. It''s probably one of my drives (judging from ata instance 3), how do I tell what drive is what instance? If I didn''t say before, thanks to all that have replied for your assistance, greatly appreciated. On Wed, Oct 20, 2010 at 2:23 PM, <Casper.Dik at sun.com> wrote:> >>Huh, I don''t actually ever recall enabling that. Perhaps that is >>connected to the message I started getting every minute recently in >>the kernel buffer, > > It''s on by default. > > You can see if it was ever enabled by using: > > ? ? ? ?zfs list -t snapshot |grep @zfs-auto > >>Oct 20 12:20:49 megatron pcplusmp: [ID 805372 kern.info] pcplusmp: id>>e >>(ata) instance 3 irq 0xf vector 0x45 ioapic 0x2 intin 0xf is bound to >>cpu 0 >>Oct 20 12:21:49 megatron pcplusmp: [ID 805372 kern.info] pcplusmp: id>>e >>(ata) instance 3 irq 0xf vector 0x45 ioapic 0x2 intin 0xf is bound to >>cpu 1 > > This sounds more like a device driver unloaded and later it is reloaded > because of some other service. > >>I just disabled it ("zfs set com.sun\:auto-snapshot=3Dfalse tank", >>correct?), will see if the log messages disappear. Did the filesystem >>kill off some snapshots or something in an effort to free up space? > > Yes, but typically it will log that. > > Casper > >-- --khd