[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [dm-devel] Re: configure MD3000



On Thu, 2009-03-12 at 15:40 -0400, Thomas Witzel wrote:
> What sort of bothers me, is that although we specified rdac, it says
> hwhandler=0 while on the web there seems to be output examples where
> it does say rdac in the multipath output.
> Also googling convinced me that some people have
> drivers/md/md-multipath-rdac.c or something like that in their kernel,
> while my ubuntu kernel and the vanilla git kernel do not have this. I
> have the feeling this is an important component of this.

>From 2.6.27 onwards, rdac has moved to scsi layer as scsi_dh_rdac. You
can do a "lsmod | grep rdac" to see if that module is inserted.

> 
> On Thu, Mar 12, 2009 at 3:38 PM, Thomas Witzel <witzel thomas gmail com> wrote:
> > How do you want me to test whether I can read/write ? I can no longer
> > mount the /dev/sd*1 partitions because now they are busy (thats since
> > I started multipath). Yes, I did execute multipath -F and multipath
> > repeatedly, the daemon is running.
> >
> > On Thu, Mar 12, 2009 at 3:31 PM, Chandra Seetharaman
> > <sekharan us ibm com> wrote:
> >> Are you able to read/write from/to the scsi device ?
> >>
> >> If yes, Did you do "multipath -F" followed by "multipath".
> >>
> >> Is the multipathd running ?
> >>
> >> chandra
> >> On Thu, 2009-03-12 at 15:25 -0400, Thomas Witzel wrote:
> >>> My /sys/block looks like this:
> >>> ls /sys/block/
> >>> dm-0  dm-3  dm-6   loop2  loop5  ram0   ram11  ram14  ram3  ram6  ram9
> >>>  sdc  sdf  sdi  sdl
> >>> dm-1  dm-4  loop0  loop3  loop6  ram1   ram12  ram15  ram4  ram7  sda
> >>>  sdd  sdg  sdj  sdm
> >>> dm-2  dm-5  loop1  loop4  loop7  ram10  ram13  ram2   ram5  ram8  sdb
> >>>  sde  sdh  sdk  sr0
> >>>
> >>> My /proc/partitions looks like this:
> >>>
> >>> major minor  #blocks  name
> >>>
> >>>    8     0   71041024 sda
> >>>    8     1   68099503 sda1
> >>>    8     2          1 sda2
> >>>    8     5    2939863 sda5
> >>>    8    16 2134896640 sdb
> >>>    8    17 2134893883 sdb1
> >>>    8    32 2134896640 sdc
> >>>    8    48 2134896640 sdd
> >>>    8    49 2134893883 sdd1
> >>>    8    64 2134896640 sde
> >>>    8    80 1222794240 sdf
> >>>    8    96      20480 sdg
> >>>    8   112 2134896640 sdh
> >>>    8   128 2134896640 sdi
> >>>    8   129 2134893883 sdi1
> >>>    8   144 2134896640 sdj
> >>>    8   160 2134896640 sdk
> >>>    8   161 2134893883 sdk1
> >>>    8   176 1222794240 sdl
> >>>    8   192      20480 sdm
> >>>  254     0 2134896640 dm-0
> >>>  254     1 2134896640 dm-1
> >>>  254     2 2134896640 dm-2
> >>>  254     3 2134896640 dm-3
> >>>  254     4 1222794240 dm-4
> >>>  254     5      20480 dm-5
> >>>  254     6      20480 dm-6
> >>>
> >>> All the sd's with a partition are working fine, I even created a test
> >>> file system at some point. I did load scsi_dh_rdac for testing at some
> >>> point, and that put the following into the syslog:
> >>> [10170.087743] sd 1:0:0:0: rdac: LUN 0 (owned)
> >>> [10170.089151] sd 1:0:0:1: rdac: LUN 1 (unowned)
> >>> [10170.089941] sd 1:0:0:2: rdac: LUN 2 (owned)
> >>> [10170.090719] sd 1:0:0:3: rdac: LUN 3 (unowned)
> >>> [10170.092112] sd 1:0:0:4: rdac: LUN 4 (owned)
> >>> [10170.092964] sd 1:0:1:0: rdac: LUN 0 (unowned)
> >>> [10170.093733] sd 1:0:1:1: rdac: LUN 1 (owned)
> >>> [10170.094507] sd 1:0:1:2: rdac: LUN 2 (unowned)
> >>> [10170.095261] sd 1:0:1:3: rdac: LUN 3 (owned)
> >>> [10170.096081] sd 1:0:1:4: rdac: LUN 4 (unowned)
> >>> [10170.096088] rdac: device handler registered
> >>>
> >>> Also loading the scsi_dh_rdac module seems to stop the [ 8996.087508]
> >>> Buffer I/O error on device dm-5, logical block 5
> >>> [ 8996.087539] Buffer I/O error on device dm-5, logical block 6
> >>> [ 8996.087548] Buffer I/O error on device dm-5, logical block 5112
> >>> [ 8996.087553] Buffer I/O error on device dm-5, logical block 5112
> >>> [ 8996.087601] Buffer I/O error on device dm-5, logical block 12
> >>> [ 8996.087606] Buffer I/O error on device dm-5, logical block 13
> >>> [ 8996.087609] Buffer I/O error on device dm-5, logical block 14
> >>> [ 8996.087612] Buffer I/O error on device dm-5, logical block 15
> >>> [ 8996.087615] Buffer I/O error on device dm-5, logical block 16
> >>> [ 8996.087618] Buffer I/O error on device dm-5, logical block 17
> >>>
> >>> type messages.
> >>>
> >>> On Thu, Mar 12, 2009 at 3:18 PM, Chandra Seetharaman
> >>> <sekharan us ibm com> wrote:
> >>> > No, you don't have to change the distro.
> >>> >
> >>> > Do you see the disks ? (the #:#:#:# makes me think that the underlying
> >>> > block devices are not configured properly). IOW, do you the sd devices
> >>> > under /sys/block ? Are you able to read/write from/to the devices ?
> >>> >
> >>> > chandra
> >>> >
> >>> > On Thu, 2009-03-12 at 15:07 -0400, Thomas Witzel wrote:
> >>> >> Thanks Chandra,
> >>> >>
> >>> >> sorry for the out of order reply, still had the digest mode on. I
> >>> >> setup the config similar to what you said
> >>> >>
> >>> >> devices {
> >>> >>         device {
> >>> >>                 vendor                  "DELL"
> >>> >>                 product                 "MD3000"
> >>> >>                 hardware_handler        "1 rdac"
> >>> >>                 path_checker            rdac
> >>> >>                 path_grouping_policy    group_by_prio
> >>> >>                 prio_callout            "/sbin/mpath_prio_rdac /dev/%n"
> >>> >>                 failback                immediate
> >>> >>                 getuid_callout          "/sbin/scsi_id -g -u /block/%n"
> >>> >>         }
> >>> >> }
> >>> >>
> >>> >>
> >>> >> and I'm still getting:
> >>> >> sdh: checker msg is "directio checker reports path is down"
> >>> >> create: 360022190009773680000214749504780 ,
> >>> >> [size=2.0T][features=0][hwhandler=0]
> >>> >> \_ round-robin 0 [prio=1][undef]
> >>> >>  \_ #:#:#:# sdb 8:16  [undef][ready]
> >>> >> \_ round-robin 0 [prio=0][undef]
> >>> >>  \_ #:#:#:# sdh 8:112 [undef][faulty]
> >>> >> sdc: checker msg is "directio checker reports path is down"
> >>> >> create: 3600221900096f27d00000ab3495047fd ,
> >>> >> [size=2.0T][features=0][hwhandler=0]
> >>> >> \_ round-robin 0 [prio=0][undef]
> >>> >>  \_ #:#:#:# sdc 8:32  [undef][faulty]
> >>> >> \_ round-robin 0 [prio=1][undef]
> >>> >>  \_ #:#:#:# sdi 8:128 [undef][ready]
> >>> >> sdj: checker msg is "directio checker reports path is down"
> >>> >> create: 360022190009773680000214a495047ce ,
> >>> >> [size=2.0T][features=0][hwhandler=0]
> >>> >> \_ round-robin 0 [prio=1][undef]
> >>> >>  \_ #:#:#:# sdd 8:48  [undef][ready]
> >>> >> \_ round-robin 0 [prio=0][undef]
> >>> >>  \_ #:#:#:# sdj 8:144 [undef][faulty]
> >>> >> sde: checker msg is "directio checker reports path is down"
> >>> >> create: 3600221900096f27d00000ab449504839 ,
> >>> >> [size=2.0T][features=0][hwhandler=0]
> >>> >> \_ round-robin 0 [prio=0][undef]
> >>> >>  \_ #:#:#:# sde 8:64  [undef][faulty]
> >>> >> \_ round-robin 0 [prio=1][undef]
> >>> >>  \_ #:#:#:# sdk 8:160 [undef][ready]
> >>> >> sdl: checker msg is "directio checker reports path is down"
> >>> >> create: 360022190009773680000214c49504806 ,
> >>> >> [size=1.1T][features=0][hwhandler=0]
> >>> >> \_ round-robin 0 [prio=1][undef]
> >>> >>  \_ #:#:#:# sdf 8:80  [undef][ready]
> >>> >> \_ round-robin 0 [prio=0][undef]
> >>> >>  \_ #:#:#:# sdl 8:176 [undef][faulty]
> >>> >> create: 360022190009773680000000000000000 ,
> >>> >> [size=20M][features=0][hwhandler=0]
> >>> >> \_ round-robin 0 [prio=1][undef]
> >>> >>  \_ #:#:#:# sdg 8:96  [undef][ready]
> >>> >> create: 3600221900096f27d0000000000000000 ,
> >>> >> [size=20M][features=0][hwhandler=0]
> >>> >> \_ round-robin 0 [prio=1][undef]
> >>> >>  \_ #:#:#:# sdm 8:192 [undef][ready]
> >>> >> libdevmapper: libdm-common.c(312): Created
> >>> >> /dev/mapper/360022190009773680000000000000000
> >>> >> libdevmapper: libdm-common.c(312): Created
> >>> >> /dev/mapper/3600221900096f27d0000000000000000
> >>> >>
> >>> >> Now this is not on RH or SLES, its an ubuntu 8.4 system. If absolutely
> >>> >> necessary I would  consider switching the distribution, but right now
> >>> >> I'd prefer to stick with what all the other machines have.
> >>> >>
> >>> >> Thanks, Thomas
> >>> >>
> >>> >> --
> >>> >> dm-devel mailing list
> >>> >> dm-devel redhat com
> >>> >> https://www.redhat.com/mailman/listinfo/dm-devel
> >>> >
> >>> > --
> >>> > dm-devel mailing list
> >>> > dm-devel redhat com
> >>> > https://www.redhat.com/mailman/listinfo/dm-devel
> >>> >
> >>>
> >>> --
> >>> dm-devel mailing list
> >>> dm-devel redhat com
> >>> https://www.redhat.com/mailman/listinfo/dm-devel
> >>
> >> --
> >> dm-devel mailing list
> >> dm-devel redhat com
> >> https://www.redhat.com/mailman/listinfo/dm-devel
> >>
> >


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]