trouble with software raid-1 on dm-multipath-devices
hi list!
i want to build a software raid-1 on top of multipath-devices and maybe need to stack an lvm on top of this (not sure yet).
some details of my trouble by doing so:
my host with 2 HBAs (emulex lp9802) is attached to 2 storage-arrays (emc symmetrix on different locations) on a dual independet fabric (1 hba on each fabric).
i have 2 paths to each symmetrix - looks like this:
SYMMETRIX1:
(0 0 0 10) sdb [ready ] (8:16)
(1 0 0 10) sdr [ready ] (65:16)
SYMMETRIX2:
(0 0 1 61) sdj [ready ] (8:144)
(1 0 1 61) sdz [ready ] (65:144)
so i just build the multipath devices:
echo "0 17677440 multipath 2 round-robin 1 0 65:16 round-robin 1 0 8:16" | dmsetup create sym1_388
...make the nodes via dmsetup mknodes etc.
to this point all is fine.
but now i want to build a mirror between the syms and *grmbl*:
mdadm -C -l mirror -n 2 /dev/md0 /dev/mapper/sym1_dev388 /dev/mapper/sym2_dev388
results in:
Jun 8 02:24:09 s01kr606 kernel: md: could not lock dm-7. Jun 8 02:24:09 s01kr606 kernel: md: error, md_import_device() returned -16
is there any other possibility or any clue to get it working??
i'm using a sles9 with sp1. in the past (kernel 2.4, sles8 w sp3) i got the same construction working with md-only (md-multipathing, stacking a md-mirror on top and finally stacking an lvm on top of the mirror) but i like the multipath implemetation with dm much more...
thanks!
-sven
Quoted 9 lines · Sven Hessenmueller
mdadm -C -l mirror -n 2
/dev/md0/dev/mapper/sym1_dev388/dev/mapper/sym2_dev388results in:
Jun 8 02:24:09 s01kr606 kernel: md: could not lock dm-7. Jun 8 02:24:09 s01kr606 kernel: md: error, md_import_device() returned -16is there any other possibility or any clue to get it working??
This should work, I test that frequently.
Are the /dev/dm-xx mounted/accessed by anything else?
(And, because of a bug, I'm hoping you're running SLES9 SP1 with a kernel updated to post-SP1, because that one was broken with DM/md stacking due to an oversight... But then, you'd have seen an oops already ;-)
SP2 will bring much improved DM multipath, too.
Actually, in the future you may want to annoy the Novell / SUSE support organization with questions about SLES9 kernels, because they are fairly different from mainline ;-)
Sincerely,
Lars Marowsky-Brée <[email protected]>
Signature · 9 lines
High Availability & Clustering SUSE Labs, Research and Development SUSE LINUX Products GmbH - A Novell Business -- Charles Darwin "Ignorance more frequently begets confidence than does knowledge" - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
Quoted 6 lines · Lars Marowsky-Bree
(And, because of a bug, I'm hoping you're running SLES9 SP1 with a kernel updated to post-SP1, because that one was broken with DM/md stacking due to an oversight... But then, you'd have seen an oops already ;-)
wish i had seen an oops, because that was it - version 2.6.5-7.155.29 works fine now.
many thanks!
Quoted 6 lines
Actually, in the future you may want to annoy the Novell / SUSE support organization with questions about SLES9 kernels, because they are fairly different from mainline ;-)
i will focus on mainline-kernels in future... ;-)
-sven
Trouble with SATA drives: "IRQ disabled"
In reply to Sven HessenmuellerHi all,
I have a machine running an array of 8 WD SATA drives, kernel 2.6.11.12 patched with libata-devel to get smart working. I'm consistently getting these errors:
Jun 28 14:53:12 governator kernel: ata5: command 0xb0 timeout, stat 0xd0 host_stat 0x0
Jun 28 14:53:12 governator kernel: ata5: status=0xd0 { Busy }
Jun 28 14:53:17 governator kernel: irq 185: nobody cared!
Jun 28 14:53:17 governator kernel:
Jun 28 14:53:17 governator kernel: Call Trace:<IRQ> <ffffffff80168e35>{__report_bad_irq+53} <ffffffff80168f0c>{note_interrupt+92}
Jun 28 14:53:17 governator kernel: <ffffffff801683c4>{__do_IRQ+628} <ffffffff8013b61e>{profile_tick+78}
Jun 28 14:53:17 governator kernel: <ffffffff80111bb8>{do_IRQ+72} <ffffffff8010f0f3>{ret_from_intr+0}
Jun 28 14:53:17 governator kernel: <EOI> <ffffffff80395d90>{unix_poll+0} <ffffffff8010cd50>{default_idle+0}
Jun 28 14:53:17 governator kernel: <ffffffff8010cd72>{default_idle+34} <ffffffff8010cdb5>{cpu_idle+53}
Jun 28 14:53:17 governator kernel: <ffffffff80552802>{start_kernel+466} <ffffffff80552243>{_sinittext+579}
Jun 28 14:53:17 governator kernel: Jun 28 14:53:17 governator kernel: handlers:
Jun 28 14:53:17 governator kernel: [<ffffffff88045000>] (nv_interrupt+0x0/0x1c0 [sata_nv])
Jun 28 14:53:17 governator kernel: [<ffffffff802f6f70>] (usb_hcd_irq+0x0/0x70)
Jun 28 14:53:17 governator kernel: Disabling IRQ #185
Jun 28 14:55:01 governator crond(pam_unix)[3349]: session opened for user root by (uid=0)
Jun 28 14:55:15 governator kernel: ata5: command 0xb0 timeout, stat 0x50 host_stat 0x0
Jun 28 14:55:25 governator kernel: ata5: command 0xb0 timeout, stat 0x50 host_stat 0x0
Jun 28 14:55:36 governator kernel: ata6: command 0xb0 timeout, stat 0x50 host_stat 0x0
Jun 28 14:55:46 governator kernel: ata6: command 0xb0 timeout, stat 0x50 host_stat 0x0Does anyone have a clue what's going on here? I remember reading somewhere that the libata patch to get smart working was not "safe" because it did not make sure a command was already executing. Could that be what's going on? I'm using smart to log the temps of the drives every 5 minutes.
Any ideas would be appreciated.
thanks,
/Patrik