[zfs-discuss] Zfs and smartctl

Felix Rubio Dalmau felix at kngnt.org
Tue Dec 15 16:08:28 EST 2015


First of all, sorry for being imprecise. I was at work and talking by heart (and there's been some days since I saw the last kernel logs). The disks are not starting and stopping every half an hour (although they used to) but more frequently: every 1-2 minutes.

After setting all -B 255 for those disks, and the link_power_management_policy to max_performance, I have observed no difference in the logs:

Dec 15 21:44:15 kernel: sd 2:0:0:0: [sdc] Starting disk
Dec 15 21:44:43 kernel: sd 3:0:0:0: [sdd] Synchronizing SCSI cache
Dec 15 21:44:43 kernel: sd 2:0:0:0: [sdc] Synchronizing SCSI cache
Dec 15 21:44:43 kernel: sd 3:0:0:0: [sdd] Stopping disk
Dec 15 21:44:43 kernel: sd 2:0:0:0: [sdc] Stopping disk
Dec 15 21:45:03 kernel: ata3: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Dec 15 21:45:03 kernel: ata3.00: configured for UDMA/133
Dec 15 21:45:03 kernel: sd 2:0:0:0: [sdc] Starting disk
Dec 15 21:45:03 kernel: ata4: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Dec 15 21:45:03 kernel: ata4.00: configured for UDMA/133
Dec 15 21:45:03 kernel: sd 3:0:0:0: [sdd] Starting disk
Dec 15 21:45:28 kernel: sd 3:0:0:0: [sdd] Synchronizing SCSI cache
Dec 15 21:45:28 kernel: sd 3:0:0:0: [sdd] Stopping disk
Dec 15 21:45:28 kernel: sd 2:0:0:0: [sdc] Synchronizing SCSI cache
Dec 15 21:45:28 kernel: sd 2:0:0:0: [sdc] Stopping disk
Dec 15 21:47:27 kernel: ata3: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Dec 15 21:47:27 kernel: ata4: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Dec 15 21:47:27 kernel: ata3.00: configured for UDMA/133
Dec 15 21:47:27 kernel: ata4.00: configured for UDMA/133
Dec 15 21:47:27 kernel: sd 2:0:0:0: [sdc] Starting disk
Dec 15 21:47:27 kernel: sd 3:0:0:0: [sdd] Starting disk
Dec 15 21:47:53 kernel: sd 2:0:0:0: [sdc] Synchronizing SCSI cache
Dec 15 21:47:53 kernel: sd 3:0:0:0: [sdd] Synchronizing SCSI cache
Dec 15 21:47:53 kernel: sd 2:0:0:0: [sdc] Stopping disk
Dec 15 21:47:53 kernel: sd 3:0:0:0: [sdd] Stopping disk

On the other hand, this is the log I obtain for sda, the disk containing the OS for the last 3 days, and started since 

Dec 13 09:44:00 kernel: sd 0:0:0:0: [sda] Synchronizing SCSI cache
Dec 13 09:44:00 kernel: sd 0:0:0:0: [sda] Stopping disk
Dec 13 09:44:00 kernel: sd 0:0:0:0: [sda] Starting disk
Dec 13 13:19:14 kernel: sd 0:0:0:0: [sda] Synchronizing SCSI cache
Dec 13 13:19:14 kernel: sd 0:0:0:0: [sda] Stopping disk
Dec 13 13:19:16 kernel: sd 0:0:0:0: [sda] Starting disk
Dec 13 16:15:56 kernel: sd 0:0:0:0: [sda] Synchronizing SCSI cache
Dec 13 16:15:56 kernel: sd 0:0:0:0: [sda] Stopping disk
Dec 13 16:15:56 kernel: sd 0:0:0:0: [sda] Starting disk
Dec 13 16:32:07 kernel: sd 0:0:0:0: [sda] Synchronizing SCSI cache
Dec 13 16:32:07 kernel: sd 0:0:0:0: [sda] Stopping disk
Dec 13 16:32:08 kernel: sd 0:0:0:0: [sda] Starting disk
Dec 13 18:35:58 kernel: sd 0:0:0:0: [sda] Synchronizing SCSI cache
Dec 13 18:35:58 kernel: sd 0:0:0:0: [sda] Stopping disk
Dec 13 18:35:58 kernel: sd 0:0:0:0: [sda] Starting disk
Dec 13 22:14:43 kernel: sd 0:0:0:0: [sda] Synchronizing SCSI cache
Dec 13 22:14:43 kernel: sd 0:0:0:0: [sda] Stopping disk
Dec 13 22:14:44 kernel: sd 0:0:0:0: [sda] Starting disk
Dec 13 22:56:44 kernel: sd 0:0:0:0: [sda] Synchronizing SCSI cache
Dec 13 22:56:44 kernel: sd 0:0:0:0: [sda] Stopping disk
Dec 13 22:56:45 kernel: sd 0:0:0:0: [sda] Starting disk
Dec 14 21:20:04 kernel: sd 0:0:0:0: [sda] Synchronizing SCSI cache
Dec 14 21:20:04 kernel: sd 0:0:0:0: [sda] Stopping disk
Dec 14 21:20:05 kernel: sd 0:0:0:0: [sda] Starting disk

To Alex: sdb is another hdd, that is mostly stopped as the data it contains is rarely accessed (the biggest, and the slowest).

Felix

On Tuesday 15 December 2015 20:08:38 Gordan Bobic via zfs-discuss wrote:
> You might want to try hdparm -B255 first. See the hdparm man page for
> details.
> Also check /sys/class/scsi_host/host*/link_power_management_policy
> 
> 
> On Tue, Dec 15, 2015 at 7:52 PM, Felix Rubio Dalmau via zfs-discuss <
> zfs-discuss at list.zfsonlinux.org> wrote:
> 
> > Ok, here is some more info: the problematic disks are sda, sdb, and sdc.
> > In sda I have the operating system, while in sd[cd] I keep the mirrored
> > data. The specs, as reported by hdparm for the disks, are:
> >
> >         OS:             debian testing, up-to-date
> >         Kernel:         compiled by me, linux 4.2.3
> >         sda:            SanDisk SDSSDP064G, APM 254, firmware 3.1.0
> >         sd[cd]:         SanDisk SDSSDP064G, APM 254, firmware 2.0.0
> >
> > Can the power saving strategies be enabled from somewhere else than the
> > disk itself (I don't know, like the kernel)? If so, what can I report to
> > help in dealing with this? OTOH: I have realised the difference in the
> > firmware versions. I will check if there is something about this issue on
> > the changelogs.
> >
> > What other information do you need?
> >
> > Thank you,
> > Felix
> >
> >
> > On Tuesday 15 December 2015 16:27:21 Gordan Bobic via zfs-discuss wrote:
> > > Looks suspiciously like a power saving measure kicking in somewhere and
> > > spinning the disk up and down. Certainly nothing to do with ZFS.
> > >
> > > On Tue, Dec 15, 2015 at 4:24 PM, Pavol Cupka via zfs-discuss <
> > > zfs-discuss at list.zfsonlinux.org> wrote:
> > >
> > > > I don't think it has to do anything with the filesystem. It sure seems
> > > > like the disk might be faulty or the the controller. Give us some more
> > > > details about the system and the disks.
> > > >
> > > > Thanks
> > > > PC
> > > >
> > > > On Tue, Dec 15, 2015 at 3:27 PM, Felix Rubio via zfs-discuss <
> > > > zfs-discuss at list.zfsonlinux.org> wrote:
> > > >
> > > >> Hi everybody,
> > > >>
> > > >>     I have two disks, using EcryptFS over ZFS, mirroring each other. I
> > > >> have set-up a short SMART self test for those disks to be run every
> > day at
> > > >> night, but the test never finishes. On the other hand I have observed
> > that,
> > > >> every half an hour, I have these messages in the log:
> > > >>
> > > >> Oct 27 02:20:38 kernel: sd 0:0:0:0: [sdb] Starting disk
> > > >> Oct 27 02:21:13 kernel: sd 0:0:0:0: [sdb] Synchronizing SCSI cache
> > > >> Oct 27 02:21:13 kernel: sd 0:0:0:0: [sdb] Stopping disk
> > > >>
> > > >>     May it have something to do with the ZFS? I am asking this on this
> > > >> list because on the same system I have another ssd disk with ext4,
> > with the
> > > >> same test finishing successfully every night. Can somebody help
> > figuring
> > > >> this out?
> > > >>
> > > >>     Thank you,
> > > >>     Felix
> > > >>
> > > >> _______________________________________________
> > > >> zfs-discuss mailing list
> > > >> zfs-discuss at list.zfsonlinux.org
> > > >> http://list.zfsonlinux.org/cgi-bin/mailman/listinfo/zfs-discuss
> > > >>
> > > >
> > > >
> > > > _______________________________________________
> > > > zfs-discuss mailing list
> > > > zfs-discuss at list.zfsonlinux.org
> > > > http://list.zfsonlinux.org/cgi-bin/mailman/listinfo/zfs-discuss
> > > >
> > > >
> >
> > _______________________________________________
> > zfs-discuss mailing list
> > zfs-discuss at list.zfsonlinux.org
> > http://list.zfsonlinux.org/cgi-bin/mailman/listinfo/zfs-discuss
> >



More information about the zfs-discuss mailing list