Read Open Error - HELP!! 3

[Voyager1]

I am constantly getting "Read Open Error - The Device is not Ready" messages that are driving me KNUTZ! Running NetWorker 5.5.3 on a Proliant server with 4 Adaptec SCSI cards. All connected to a StorageTek 9714 with 4 DLT 7000 drives.

We have had the hardware checked by StorageTek and they have certified that all is OK.

Unfortunately, this is all happening at remote sites, so I don't have direct access to the STK device.

Any help would be greatly appreciated!! X-) Bill
CNE, MCSE, CCNA, CCA, MCT, CCI

"I'd much rather be lucky than good"

 

[cfowler]

You could attempt to recreate your jukebox / autochanger using the jbconfig command. You will have to go through the gui to delete the existing device, the shut down all services on the networker server, restart the services, run jbconfig and then inventory the jukebox again.

Hope this helps,

cfowler

 

[Palindrome]

Bill,

What is your backup server? Is it a Compaq Proliant DL380? I'm currently investigating the same type of problem you are having. I have successfully loaded Legato 6.01 on a Compaq 2500 with no problems. Please let me know. THX

 

[Voyager1]

A little new information since I posted this question. First to cfowler, thanks for the input, but I've "been-there done-that" several times. That was the first thing that crossed my mind, too. To Palindrome (that's kinda funny), my server is a Compaq Proliant 5000, dual 333s with a gig of RAM.

According to Quantum (makers of the DLT7000), there is an "error threshhold" in Legato that needs to be increased. The Legato interface calls it "consecutive errors" and of this threshold gets crossed the "read open error" will surface. The only way to reset the error is to clear the drive and reboot the server. Their recommendation was to increase this setting to at least 100. That, plus upgrading to 6.0.1 seems to have resolved the problem.

Thanks for the input... - Bill

"You can get anything you want out of life, if you'll just help enough other people get what they want" - Zig Ziglar

 

[cfowler]

Voyager1,

Yeah, I hear you there. I'm going through it again. Don't know what happened but for the past 2 weeks, we've been having to perform disaster recovery on our server quite a bit more than is realistic. Something's happened to the hardware config and I can't find it right now. However, your latest post certainly does provide some bright input.

Thanks for sharing! B-)

cfowler

 

[x2shy]

Ohhhh. DL380. My too and I am having the same problem....
Something afoot me thinks. Storagetek suggested it might be an issue with the the internal directory structure on one of the tapes that is throwing the drive out of line. Not too sure about that but it makes _half_ sense.
btw alrady at 6.0.1 and gettting the error.

 

[Voyager1]

Here's an interesting tidbit: I traveled to one of our remote sites to reconfig the system and have a meeting with our software (Legato) vendor. He brought along an engineer from StorageTek who said that having 20 (20!) drive failures in a 4-drive unit over the course of a year is not unusual. Can you believe that?!?

His recommendation was, or course, to upgrade to the DLT8000 drives ($$).

And so it goes.......LOL - Bill

"You can get anything you want out of life, if you'll just help enough other people get what they want" - Zig Ziglar

 

[Palindrome]

X2shy,

We have narrowed the problem down to the Host Bus Adapter that was purchased with the DL380. We tried an Emulex card and still had the same problem. Q-Logic 2200 HBA fixed the problem for us. We tested backups and restores and everything seems fine. We are using a SAN environment and am not sure what you are using for a network. Perhaps changing the SCSI or Fibre card (If using a SAN) will fix the problem. Let me know if this help. Thx

 

[Voyager1]

We're not doing anything as sexy as a SAN 9although we should). We're simply client/server with a pretty healthy Compaq Proliant server, through 4 Adaptec differential SCSIs to the StorageTek. Yawn... - Bill

"You can get anything you want out of life, if you'll just help enough other people get what they want" - Zig Ziglar

 

[cfowler]

Wanted to say it appears as though we have our problem solved. After numerous attempts with config alterations, load_sleep, eject_sleep settings and others, it appears as thought the 9840 devices or it could be the 9740 Autochanger itself, does not like any of the SCSI device drivers for the 2944UW other than what are supplied within Winnt Server 4.0.

Installing the NT 4.0 drivers in this case, solved the problem. At least at this point it's smooth sailing again.

Great board!

cfowler

 

[frende]

I had the Read Open errors a couple of months ago and found out from Legato that it was actually a problem with the SCSI chain. We replaced the firmware and reset everything and when the drives where rebuilt in Networker we dident have the problem any more.

Hope this helps.

 

[Barks]

I've seen this message a lot too, but the cause of it in my environment was hardware cleaning being turned on. When the tape library loads a cleaning tape and at the same time Networker trys to load a tape for a backup, Networker gets confused. I'm cleaning my drives manually now and the read open errors have now gone away.
Kent

 

[Matts]

This may come a little late, but... Historically Compaq servers never really worked well with SCSI cards from other vendors, especially Adaptec. Now the funny part: Compaq OEM's an Adaptec card for their Ultra3 64-bit 66MHz card, although modified with a Compaq firmware and Compaq drivers.

So maybe it cold be an idea to try a Compaq SCSI card...?

Good luck!

 

[techtipp]

Hi!
We are running Networker 6.0 on Solaris 2.8 / sun420r and exabyte mammoth-ii. (/dev/rmt/0)
Backup is no problem,but recovering files getting always this error message :

--
media info: can not read record 0 of file 2 on 8mm Mammoth-2 tape
--
any idea ?

hbellenberg

 

[spectre]

Myself, having only become aware of an issue due to ndmp configuration, you could also try the load sleep attribute plus the other sleep attributes of the jukebox resource. This would only be relevant if your errors were at job start, not mid-stream.

 

[cptb]

Read open error is generated when Networker thinks it has a tape mounted in a drive when it's actually empty. This type of error doesn't (normally anyway) generate any consecutive errors why you don't hav to increase that number. The environment I've seen this in is mostly SANs and my oppinion is that it has to do with SCSI command delays within the SAN.

 

[packageman]

Hmmm. We have 2 sites with that same exact hw: SUn Netra t1s with Storagetek 9730s, SCSI cards, SOlaris 8, 6.1.1 build 238. The "Read open" only happens at one site. It causes the drive to eat the tape and not let go. I have to powercycle the library and reboot the server to remove the tape.