Hi, The reason zfs is saying that the available is larger is because in Zfs the size of the pool is always available to the all the zfs filesystems that reside in the pool. Setting a reservation will gaurntee that the reservation size is "reserved" for the filesystem/volume but you can change that on the fly.
You can see that if you create another filsystem within the pool that the reservation in use by your volume will have be deducted from the available size. Like below: r...@testfs create -V 10g testpool/test r...@testfs get all testpool NAME PROPERTY VALUE SOURCE testpool type filesystem - testpool creation Wed Feb 11 13:17 2009 - testpool used 10.1G - testpool available 124G - testpool referenced 100M - testpool compressratio 1.00x - testpool mounted yes - Here the available is 124g as the volume has been set to 10g from a pool of 134g. If we set a reservation like this r...@test1 set reservation=10g testpool/test r...@test1 zfs get all testpool/test NAME PROPERTY VALUE SOURCE testpool/test type volume - testpool/test creation Tue Mar 3 10:13 2009 - testpool/test used 10G - testpool/test available 134G - testpool/test referenced 16K - testpool/test compressratio 1.00x - We can see that the available is now 134g, which is the avilable size of the rest of the pool + the 10g reservation that we have set. So in theory this volume can grow to the complete size of the pool. So if we have a look at the availble space now in the pool we see r...@test1# zfs get all testpool NAME PROPERTY VALUE SOURCE testpool type filesystem - testpool creation Wed Feb 11 13:17 2009 - testpool used 10.1G - testpool available 124G - testpool referenced 100M - testpool compressratio 1.00x - testpool mounted yes - 124g with 10g used to account for the size of the volume ! So if we now create another filesystem like this r...@test1# zfs create testpool/test3 r...@test1# zfs get all testpool/test3 NAME PROPERTY VALUE SOURCE testpool/test3 type filesystem - testpool/test3 creation Tue Mar 3 10:19 2009 - testpool/test3 used 18K - testpool/test3 available 124G - testpool/test3 referenced 18K - testpool/test3 compressratio 1.00x - testpool/test3 mounted yes - We see that the total amount available to the filesystem is the amount of the space in the pool minus the 10g reservation. Lets set the reservation to something bigger. r...@test1# zfs set volsize=100g testpool/test r...@test1# zfs set reservation=100g testpool/test r...@test1# zfs get all testpool/test NAME PROPERTY VALUE SOURCE testpool/test type volume - testpool/test creation Tue Mar 3 10:13 2009 - testpool/test used 100G - testpool/test available 134G - testpool/test referenced 16K - So the available is still 134G, which is the rest of the pool + the reservation set. r...@test1# zfs get all testpool NAME PROPERTY VALUE SOURCE testpool type filesystem - testpool creation Wed Feb 11 13:17 2009 - testpool used 100G - testpool available 33.8G - testpool referenced 100M - testpool compressratio 1.00x - testpool mounted yes - The pool however now only has 33.8G left, which should be the same for all the other filesystems in the pool. Hope that helps. -----Original Message----- From: zfs-discuss-boun...@opensolaris.org [mailto:zfs-discuss-boun...@opensolaris.org]on Behalf Of Lars-Gunnar Persson Sent: 03 March 2009 07:11 To: Richard Elling Cc: zfs-discuss@opensolaris.org Subject: Re: [zfs-discuss] ZFS volume corrupted? ************************************* This e-mail has been received by the Revenue Internet e-mail service. (IP) ************************************* I've turned off iSCSI sharing at the moment. My first question is: how can zfs report available is larger than reservation on a zfs volume? I also know that used mshould be larger than 22.5 K. Isn't this strange? Lars-Gunnar Persson Den 3. mars. 2009 kl. 00.38 skrev Richard Elling <richard.ell...@gmail.com >: > Lars-Gunnar Persson wrote: >> Hey to everyone on this mailing list (since this is my first post)! > > Welcome! > >> >> We've a Sun Fire X4100 M2 server running Solaris 10 u6 and after >> some system work this weekend we have a problem with only one ZFS >> volume. >> >> We have a pool called /Data with many file systems and two volumes. >> The status of my zpool is: >> >> -bash-3.00$ zpool status >> pool: Data >> state: ONLINE >> scrub: scrub in progress, 5.99% done, 13h38m to go >> config: >> >> NAME STATE READ WRITE CKSUM >> Data ONLINE 0 0 0 >> c4t5000402001FC442Cd0 ONLINE 0 0 0 >> >> errors: No known data errors >> >> >> Yesterday I started the scrub process because I read that was a >> smart thing to do after a zpool export and zpool import procedure. >> I did this because I wanted to move the zpool to another OS >> installation but changed my mind and did a zpool import on the same >> OS as I did an export. >> >> After checking as much information as I could find on the web, I >> was advised to to run the zpool scrub after an import. >> >> Well, the problem now is that one volume in this zpool is not >> working. I've shared it via iscsi to a Linux host (all of this was >> working on Friday). The Linux host reports that it can't find a >> partition table. Here is the log from the Linux host: >> >> Mar 2 11:09:36 eva kernel: SCSI device sdb: 524288000 512-byte hdwr >> sectors (268435 MB) >> Mar 2 11:09:36 eva kernel: SCSI device sdb: drive cache: write >> through >> Mar 2 11:09:36 eva kernel: SCSI device sdb: 524288000 512-byte hdwr >> sectors (268435 MB) >> Mar 2 11:09:37 eva kernel: SCSI device sdb: drive cache: write >> through >> Mar 2 11:09:37 eva kernel: sdb: unknown partition table >> Mar 2 11:09:37 eva kernel: Attached scsi disk sdb at scsi28, >> channel 0, id 0, lun 0 >> >> >> So I checked the status on my Solaris server and I found this >> information a bit strange;: >> >> -bash-3.00$ zfs list Data/subversion1 >> NAME USED AVAIL REFER MOUNTPOINT >> Data/subversion1 22.5K 519G 22.5K - >> >> How can it bed 519GB available on a volume that is 250GB in size? >> Here are more details: >> >> -bash-3.00$ zfs get all Data/subversion1 >> NAME PROPERTY VALUE SOURCE >> Data/subversion1 type volume - >> Data/subversion1 creation Wed Apr 2 9:06 2008 - >> Data/subversion1 used 22.5K - >> Data/subversion1 available 519G - >> Data/subversion1 referenced 22.5K - >> Data/subversion1 compressratio 1.00x - >> Data/subversion1 reservation 250G local >> Data/subversion1 volsize 250G - >> Data/subversion1 volblocksize 8K - >> Data/subversion1 checksum on default >> Data/subversion1 compression off default >> Data/subversion1 readonly off default >> Data/subversion1 shareiscsi off local > > It does not appear that Data/subversion1 is being shared via iscsi? > -- richard > > _______________________________________________ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss ************************ This message has been delivered to the Internet by the Revenue Internet e-mail service (OP) ************************* _______________________________________________ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
