Topic: Seagate STT220000A install probs

[Mike Schultz]

I think I have something similar to Frank's problem, but I can't say for sure since I'm not using the same drive.

Based on Darrell's HOWTO I dropped in a Seagate STT220000A ATAPI tape drive and everything appeared to go smoothly.  However, whenever a drive access is attempted, I end up with these logs:

Jul 24 17:59:26 e-smith kernel: st0: Error with sense data: [valid=0] Info fld=0
x0, Current st09:00: sense key Illegal Request
Jul 24 17:59:26 e-smith kernel: Additional sense indicates Invalid command operation code

The flexbackup log always indicates that the last level 0 backup was the epoch.  Also, flexbackup -list only every shows the contents of /boot, but never /, even after

mt -f /dev/nst0 fsf 1.

flexbackup -toc always complains that the tape doesn't have an index (is this normal?).

Attempting to do a selective backup using

flexbackup -extract -files /root/flexbackup/extract-list

always complains that the desired file(s) was not found on the tape.  The complete path to the file is provided in the extract list.

So, I'm not convinced this drive is doing anything but whirring around pretending to protect my data.  What else can I do to verify if there is anything whatsoever on the tape?

TIA
Mike Schultz

[Gordon]

Hi Mike
I have found that I need to change the backup device from nst0 to nht0 (the default name for an IDE tape in Red Hat).
There is a configuration DB command but it has not worked for me so I have used MC to locate all the references in the configuration files to nst0 and changed them manually to nht0. you also need to use the right size/type of tapes in the drive (ie) TR5
Hope this helps
Gordon

[Mike Schultz]

After changing tape device references to /dev/nht0, the log errors were eliminated, but I am still unable to restore files.  After the backup has completed, I attempt to get a file from the tape using:

/bin/mt -f /dev/nht0 fsf 1 (I tried fsf 0 as well)
/usr/bin/flexbackup -extract -files /root/flexbackup/extract-list

where my extract-list file contains the full path of the file I wish to restore (in this case, /boot/boot.b).  I get:

|------------------------------------------------
| Reading from current tape position
|------------------------------------------------
At block 1.
|------------------------------------------------
| Extracting files listed in /root/flexbackup/extract-list
| buffer -m 3m -s 10k -u 100 -t -p 75 -B -i /dev/nht0 | gzip -dq | restore -x \
|  -o -v -b 10 -f - boot/boot.b
|------------------------------------------------
Kilobytes Out 20

gzip: stdin: not in gzip format
restore: Tape is not a dump tape
Verify tape and initialize maps
End-of-tape encountered
|------------------------------------------------
At block 3.
|------------------------------------------------

It looks as though gzip dies in this pipe and hence restore's input is hosed as well.  Gordon, how did you verify that your backups are actually working?  Curious.
Thanks,
MIke

[Gordon]

Mike
I have just gone through a similar thing and found to restore I needed to start the tape in a different position.
From the command enviroment try typing
flexckbackup -list -num 2
If it does not work try other nubers until your screen fills with the list of backuped data. After that you will need to run a manual restore as I could not find a way to use the browser system.
I hope this helps you find your data.
Regards
Gordon

[Mike Schultz]

Hi Gordon,

Thanks for the tip.   I am starting to get a better feel for how flexbackup is put together now.  Using

flexbackup -num 2 -list

I am able to get the list of (supposedly) backed up data as you indicate.  However, try as I may, if I try to restore a file using

flexbackup -num 2 -extract -files /root/flexbackup/extract-list

all of the directory structure gets created, the tape advances to file number two, and then it spins essentially forever without coming back.  Should it take more than an hour to retrieve a single file?

If this is working for you, then maybe I have mis-configured something earlier on (like a missed reference to nst0 maybe).  The only ref to nst0 that I could find was in flexbackup.conf.   Maybe the nst0 is still lurking in the db somewhere.  I notice in the logs that st0 is still being grabbed on tape access.  Perhaps this is a clue.

Thanks for your help,
Mike

[Jonathan Chan]

I have the exact same problem.  The directory is created and the
tape is keep spining for 15mins.  But no file is restored.  
Can someone help both of us here? anyone try to test their restore?

Thanks,
Jonathan

[Gordon]

Mike
I have restored using the -num 2 option but it did load all the files into the current directory with the usual tree. This seems to be a problem when the HDD is SCSI or hardware Raid (like 3ware) and IDE Tape. Still would like to find a template modification to instruct the system to reliably use nht0 in place of nst0. But would suggest you may have missed an nst0 reference.
Gordon

[Mike Schultz]

Everything is now working well with nht0.  The restore works, it just takes forever (like, 1 hour).  Backing up a small subset of / is obviously much faster and files are easily recovered in a timely manner.  Once I determined that the data was on the tape (using the smaller dataset), I had more patience to retrieve a single file from the full image.  Originally it had seemed like forever.

Thanks for the suggestions,
Mike Schultz

[Bruce Friedman]

I have seen the exact same problems in SME V5 that Mike reported in July.  I am running on a system with no SCSI - only IDE hd on hda, Seagate STT220000A on secondary IDE as master (hdc) and IDE CD-ROM on secondary IDE as slave (hdd).

The problem is that flexbackup does not seem to create an index, and /var/log/messages issues errors with sense data and invalid command operations code - same as Mike reported.

Command "flexbackup -newtape"
gives:
Oct 28 12:41:04 popeye kernel: st0: Error with sense data: [valid=0] Info fld=0x
0, Current st09:00: sense key Illegal Request
Oct 28 12:41:04 popeye kernel: Additional sense indicates Invalid command operat
ion code
Oct 28 12:41:04 popeye kernel: st0: Error with sense data: [valid=0] Info fld=0x
0, Current st09:00: sense key Illegal Request
Oct 28 12:41:04 popeye kernel: Additional sense indicates Invalid field in param

I have tried the recommended:
/sbin/e-smith/db configuration setprop backup Device /dev/nht0
/sbin/e-smith/signal-event conf-backup

This changed the error codes on "flexbackup -newtape" to:
Oct 28 12:24:19 popeye kernel: ide-tape: hdc <-> ht0, 1000KBps, 6*54kB buffer, 5
400kB pipeline, 130ms tDSC
Oct 28 12:25:15 popeye kernel: ide-tape: ht0: I/O error, pc =  a, key =  5, asc
= 80, ascq = 81
Oct 28 12:25:15 popeye kernel: ide-tape: ht0: I/O error, pc =  a, key =  5, asc
= 80, ascq = 81
Oct 28 12:25:15 popeye kernel: ide-tape: ht0: I/O error, pc = 10, key =  5, asc
= 80, ascq = 81
Oct 28 12:25:15 popeye kernel: ide-tape: Couldn't write a filemark

In both cases, flexbackup seems to say its OK when you issue the command, in spite of the errors in /var/log/messages

However, also in both cases, "flexbackup -toc" says the tape does not have an index after issuing the "flexbackup -newtape".

Note that I needed to reboot the server after changing the configuration from /dev/nst0 to /dev/nht0 and vice versa - otherwise the device is not found.

When I followed Darrel May's advice on how to read the tape:
mt -f /dev/nst0 rewind
mt -f /dev/nst0 fsf 1
flexbackup -list

I can see the files on the tape from a backup.

Does this mean the errors can be ignored?

Does anyone have any ideas on what to try next - I'm a little confused since the manual states that the Seagate STT220000A is the only supported IDE tape drive - and hardware-wise there isn't anything unusual about my system.

[Daniel]

I get these also:

ht0: I/O error, pc = a, key = 5, asc

but it backs up and restores OK.