Problems with a Ultrium Scalable tape library 3583

Can you read or write to a tape using the 'mt' command? If this not work, you might have a problem with your OS sending information to the devices, or even a driver problem.

Once I spent a whole week trying to make the inquire command works in an AIX (in a SAN), and finally found that even if it does not work, Networker would work fine.

Hi,
I can read and write from/to tapes, in terms of AIX all the devices are fine.

Well, I have no idea at the moment. If it was me, I would try some reboots and see if jbconfig detects the library, don't worry about inquire command if it does not work.

It's been some time since I had my hands on AIX, but I remember few tips... One was to set up lus (same as on Solaris).

Do something like lsdev -C | grep -i lus and make sure that lus is in available set (if not then cfmgr -i lus or mkdev -l lus should do it). As on Solaris so I think on AIX lus.cong should be set to your HBA (on Solaris you get warning messages after st init in /var/adm/messages so you will see if you have to define HBA or not). Try to look for file called lus.conf if it exists on your system (I know it does for Solaris so it should be for AIX as lus used to be used there too). Without it you won't see all devices with inquire, but you will still be able to run jbconfig. If you don't see library then that would be abother issue. Do you see the library with inquire? (and inquire -l)

Also, in the old days (and I believe now) smc had to be disabled (eg. smc0) because otherwise IBM driver would open device before Legato one (lus) and then you would get big trouble - but this is valid only for SCSI world. If your library is also connected via fiber, then do not remove smc as smc driver needs to be in "Defined" state as opposite to "Available". (actually, looking at your output it looks like it is fiber connected, but it is available).

There used to be in old knowledgebase article about this too. If I find it I will post it.

OK, here we go:
- Support Note 50067
- Support Note 68798

I hope this helps a bit.

If you have no SCSI devices at all on the system that could be an issue as well according to that note In that case you have kind of workaround which should quickly give you an idea if that was the cause or not. Nevertheless, smc suggestion still applies...

Thanks for your help, I will see if these two support notes can help me, but I think my problem is explained on support note 66828.

Hi,

I have a similar configuration of cferrao, networker server 7.2.1 with a IBM 3583 tape library connected to a SAN.

My issue is, I don't know why, but the backup speed is of 12Mb/s overall!!!! This means that when I start a client backup, this goes to a drive (drive rmt0) and starts backing up at 12Mb/s. If a second backup starts from a different client, this goes to a different tape drive (rmt1) and the speed is like 900Kb/s. If a third one starts, it goes to a different tape and the speed is like 5kb/s.

I've been trying to figure it out what is the issue in here... but I haven't found any clues.

Can someone give some hints...?

Is this network backup? If your overall speed would 12MB (or more precisly 11MB/s) then perhaps your bottleneck is 100MBit setting on the NIC (on the storage node which I assume is your backup server). Could also port setting on switch of course. So check that first. To verify that you could try to run ftp from the same hosts which you use for backup and bove described scenario.

That make sense... thanks. Yes, is a network backup and the speed of the NIC is 100Mbit on the server, I need to look for that on the switch. That will answer why the 11MB/s... but (there is always a but!!!), why the first client to start the backup keeps the highest rate!!!

I first thought that was some issue of priority, but all clients have the same priority. I would expect to see the same rates among all the backups, but isn't the case.

From here, what I need to do is to have another NIC in the backup server and try to balance de backups among the NICs