Re: Restore Problems 2.4.4 - Unexpected EOF
Thanks for the reply Jon and sorry for the delay in responding. On Wed, 18 Jun 2003, Jon LaBadie wrote: > > I've been using amanda 2.4.1 on Solaris 2.5.1 with a DLT4000 for several > > years without many problems. I've recently upgraded to Solaris 9, amanda > > 2.4.4 and a Sun L280 (6-slot DLT7000). Most of the FS's are VxFS with a > > few UFS. I'm just doing some test restores from the first night production > > backups and have run into a problem I can't resolve. > > IIRC, VxFS has its own dump/restore programs as do ufs type file systems. > > How are you handling the selection of one pair vs the other. I don't > recall anything in amanda that allows varying dump/restore programs > to be used. Maybe I missed something? Amanda does do this herself. I've been dumping a mixture of vxfs and ufs FS's since I started using amanda. vxdump is automatically chosen for vxfs and ufsdump for ufs. > > The restores from VxFS partitions (via amrecover) work fine. But when I > > try to restore from one of the UFS partitions it dies. I select the date, > > host and FS, add a file to restore and extract. It ask me to confirm and > > then load the correct tape. The tape is already loaded and rewound. As > > soon as I type Y for the final (second) confirm it just returnes the > > shell prompt. There are no errors in amrecover.*.debug but I get the > > following at the end of amindexd.*.debug : > > The debuglogs should show what restore program was being invoked. > Was it ufsrestore? The amrecover.*.debug log reports sucessful connection to the index server, a listing of the files, multiple : add_file: Pondering ditem->path="/etc/vfstab.prevm" and then : amrecover: stream_client_privileged: connected to 131.227.86.16.10083 amrecover: stream_client_privileged: our side is 0.0.0.0.800 amrecover: try_socksize: receive buffer size is 65536 amindexd.*.debug reports similar success wrt connections and file listings but ends with : amindexd: time 23.641: < 200 Opaque list of /etc amindexd: time 52.703: ? unexpected EOF amindexd: time 52.703: removing index file: /var/opt/amanda/ccsrB/index/machine.domain.com/_/20030623_0 amindexd: time 52.703: pid 11306 finish time Tue Jun 24 10:50:46 2003 I can see no mention in any of the debug files regarding the restore program which leads me to believe that it isn't getting that far. Many thanks. Cheers, Adam
Restore Problems 2.4.4 - Unexpected EOF
Hi All, I've been using amanda 2.4.1 on Solaris 2.5.1 with a DLT4000 for several years without many problems. I've recently upgraded to Solaris 9, amanda 2.4.4 and a Sun L280 (6-slot DLT7000). Most of the FS's are VxFS with a few UFS. I'm just doing some test restores from the first night production backups and have run into a problem I can't resolve. The restores from VxFS partitions (via amrecover) work fine. But when I try to restore from one of the UFS partitions it dies. I select the date, host and FS, add a file to restore and extract. It ask me to confirm and then load the correct tape. The tape is already loaded and rewound. As soon as I type Y for the final (second) confirm it just returnes the shell prompt. There are no errors in amrecover.*.debug but I get the following at the end of amindexd.*.debug : amindexd: time 14.162: < 201- 2003-06-16 0 ccsrB01 2 /adm/vold.log amindexd: time 14.162: < 201- 2003-06-16 0 ccsrB01 2 /adm/wtmpx amindexd: time 14.162: < 200 Opaque list of /adm amindexd: time 28.670: ? unexpected EOF amindexd: time 28.670: removing index file: /var/opt/amanda/ccsrB/index/machine.domain.com/_var/20030616_0 amindexd: time 28.670: pid 26807 finish time Tue Jun 17 15:51:17 2003 (names changed to protect the innocent) Can't see anything similar in the archives and google didn't produce anything interesting. Also I'm a bit confused about what amrecover_changer should point to in amanda.conf. I currently have it set to the same string as changerdev. Is this correct ? Many thanks in advance. Cheers, Adam
Re: Setting up Sun L280
Thanks once again for your help John. You provided plenty of food for thought. On Wed, 2 May 2001, John R. Jackson wrote: > >So far I have tried the following : > >- built and installed sst form the 2.4.2p2 source. I have a /dev/rsst1 and > >running sstest '/dev/rsst1 rew' I get 'Device rewound, status = > >0x0'. ... > > You tried to rewind a tape changer? I'm surprised it didn't stick its > tongue out at you and drop all the tapes on the floor. :-) I should have been paying a bit more attention ;) > You might grab: > > ftp://gandalf.cc.purdue.edu/pub/amanda/atl_op Got it. > That's a little test program I use to exercise my ATL robot through the > sst interface. After you chmod +x it, try these things: > > atl_op -h # to see the help message > atl_op -f /dev/rsst1 -v inquiry > > But there are other problems you need to look at first ... > > >- I have built the latest mtx from sourceforge : > > # mtx -f /dev/rsst1 inquiry > > Product Type: Disk Drive > > Ummm, that would seem to be a very bad thing. You're talking to a disk > drive, not your changer. :-) > > That, in turn, implies /dev/rsst1 is not pointing where you want it to. > > Were any other /dev/rsst* devices created? What does rsst1 point to? Only /dev/rsst1 was created which as you say points to the drive not the changer. I tried to change the sst.conf file as follows : name="sst" parent="/pci@1f,2000/scsi@1/sst@1,0:character" target=1 lun=0;A but this caused a kernel panic ! I created the link by hand : ln -s ../devices/pci@1f,2000/scsi@1/sst@1,0:character rsst2 and now I can : # ./atl_op -f /dev/rsst2 inquiry atl_op: device type"CHANGER" atl_op: removeable atl_op: vendor "HP " atl_op: product"C6280-7000 " atl_op: revision_level "2.10" but still get : # ./mtx -f /dev/rsst2 status mtx: Request Sense: Long Report=yes mtx: Request Sense: Valid Residual=no mtx: Request Sense: Error Code=0 (Unknown?!) mtx: Request Sense: Sense Key=No Sense mtx: Request Sense: FileMark=no mtx: Request Sense: EOM=no mtx: Request Sense: ILI=no mtx: Request Sense: Additional Sense Code = 00 mtx: Request Sense: Additional Sense Qualifier = 00 mtx: Request Sense: BPV=no mtx: Request Sense: Error in CDB=no mtx: Request Sense: SKSV=no READ ELEMENT STATUS Command Failed but doing an inquiry produces the same results as atl_op and inventory causes some changer activity but nothing to stdin. The following gets logged to /var/adm/messages : unix: WARNING: /pci@1f,2000/scsi@1/sst@1,0 (sst2) Error for Command: Error Level: Fatal unix: Requested Block: 0 Error Block: 0 unix: Vendor: HP Serial Number: 99255B9697 unix: Sense Key: Illegal Request unix: ASC: 0x24 (invalid field in cdb), ASCQ: 0x0, FRU: 0x0 > What do you get if you do this: > > find /devices -name 'sst*' # find /devices -name 'sst*' /devices/pci@1f,4000/scsi@3/sst@1,0:character /devices/pci@1f,2000/scsi@1/sst@1,0:character The top one is an internal disk drive (I think) and the bottom one is the changer. > What did the system reboot say (e.g. /var/adm/messages) when it scanned > the devices, in particular about your changer or the sst devices? unix: sst0:found Direct Access device at tgt1, lun0 unix: sst0:Vendor/Product ID = IBM DDRS34560SUN4.2G unix: sst0 at glm0: unix: target 1 lun 0 unix: sst0 is /pci@1f,4000/scsi@3/sst@1,0 unix: sst1:No response from target 1, lun 0 unix: sst2:found Changer device at tgt1, lun0 unix: sst2:Vendor/Product ID = HP C6280-7000 unix: sst2 at glm2: unix: target 1 lun unix: sst2 is /pci@1f,2000/scsi@1/sst@1,0 unix: sst3:No response from target 1, lun 0 What do I need to do ?? Thanks once again John. Cheers, Adam.
Setting up Sun L280
Hi, I've been using amanda (2.4.1p1) for quite a while with complete sucess. I have recently purchased two Sun L280 changers (6 slots, 1 DLT 700 drive) to replace my single slot units. I have decided to upgrade to 2.4.2p2 running on SPARC Solaris 7 and am having real problems getting the changer to work. I have been through the archives, faq's etc.. and although there are numerous posting of a similar vein I haven't found any relevant resolutions. So far I have tried the following : - built and installed sst form the 2.4.2p2 source. I have a /dev/rsst1 and running sstest '/dev/rsst1 rew' I get 'Device rewound, status = 0x0'. There are no reported errors with any of the test commands but I am not sure whether this confirms that all is OK. (I have also tried sg from SUN which I couldn't get to install and stctl but found out that this doesn't support the L280) - I have built the latest mtx from sourceforge : # mtx -f /dev/rsst1 inquiry Product Type: Disk Drive Vendor ID: 'IBM ' Product ID: 'DDRS34560SUN4.2G' Revision: 'S98E' Attached Changer: NoA # mtx -f /dev/rsst1 status mtx: Request Sense: Long Report=yes mtx: Request Sense: Valid Residual=no mtx: Request Sense: Error Code=0 (Unknown?!) mtx: Request Sense: Sense Key=No Sense mtx: Request Sense: FileMark=no mtx: Request Sense: EOM=no mtx: Request Sense: ILI=no mtx: Request Sense: Additional Sense Code = 00 mtx: Request Sense: Additional Sense Qualifier = 00 mtx: Request Sense: BPV=no mtx: Request Sense: Error in CDB=no mtx: Request Sense: SKSV=no READ ELEMENT STATUS Command Failed I've searched the mtx archives but found no resolutions. I'm not sure if I am going along the right line here - am I using the appropiate SCSI drive / glue script combo. I've spent a lot of time on this so far and just wanted to confirm that I am indeed going in the right direction before I continue. Has anyone else got this combination working. Am I doing the right thing. What should I do next. Any help or advise would be grately received. Many thanks, Adam.