Jump to content

VXA Problems - Solved or Not?


Recommended Posts

As I've noted recently, the thread about problems with Retrospect 5/OS X and the Ecrix VXA-1 drive hasn't seen much activity lately.

 

 

 

 

To those who were having problems:

 

 

 

 

1. Have you solved them and, if so, how?

 

 

 

or

 

 

 

2. Have you given up?

 

 

 

 

To the Dantz moderators:

 

 

 

 

1. Is this problem receiving any attention from Dantz?

 

 

 

I don't expect an immediate solution, and I'd be willing to buy a new SCSI card to replace my Adaptec 2906 if that's what it takes, but I don't think it's unreasonable to expect some sort of response or guidance.

 

 

 

-waveman

Link to comment
Share on other sites

I have an ATTO card with a VXA-1 wide SCSI drive. It doesn't hang, but it does pause, backspace and resume as has been reported by others. This causes a dramatic drop in transfer rates from 180+ MB/s to 105 MB/s when compared to the same hardware setup running 4.3 and 5.0 Preview.

 

 

 

In situtations like this, where there are multiple variables and users report different symptoms, it is usually wise to consider the possibility that there is more than one problem required to provide a complete explanation. There seems to be a very strong correlation between VXA-1 drives, Retro 5.0 (but not Preview) and the pausing syndrome. There also seems to be a strong correlations between the time sequence of hangs following pause-like syndromes. Additionally those with Adaptec cards or Firewire cards tend to see hangs, whereas it appears those with ATTO cards do not. But at least one ATTO card user (me) sees the pauses.

 

 

 

I would almost bet there are two or more independent problems here, that couple to cause the hang. Dantz has modified the tape drive handling since 4.3/5.0 Preview, at least with respect to VXA-1. You can also demonstrate this different behavior at the conclusion of a volume or client backup. The only time the drive would pause this way with the old software was handling a real tape error.

 

 

 

I have established that the transfer rate drop is entirely caused by the pause syndrome through measurement and calculation. This pausing is extremely regular and predictable, and appears to have no correlation with brand new or well worn media, or with a freshly cleaned or disgustingly dirty VXA drive. I can't imagine why Dantz did this because the VXA backup and recovery tape handling has been ultra reliable, and very efficient for me for nearly two years with 4.3 and then 5.0 Preview.

 

 

 

Those with hangs often seem to discuss the pausing problem, and/or note the failure occurs at the same boundary points that trigger the pauses. Only Adaptec and Firewire users report the hangs. So one might conjecture that something about the pausing triggers another problem for these interfaces that ultimately results in the hang.

 

 

 

Who knows, maybe something horrible or illegal is done by Retro on these roughly 200 MB boundaries that causes different responses from different hardware. Maybe there is nothing wrong with the Adaptec/Firewire interfaces because they are responding to some undefined scenario. On the other hand there may be real hardware bugs in these interfaces that cause incorrect reactions to the pausing sequence, whereas the ATTO cards work. Dantz should know what's going on, because they HAVE made a change.

 

 

 

What is obvious is that Dantz is non-responsive to a set of fairly widely reported problems and symptoms.

 

 

 

Dantz: if you can't reporduce these problems and/or need our help, please ask. I have always provided you with as much information as I could gather, including during the Preview phase. I believe others want to help, too. Your new support policy seems well crafted to insulate you from solving critical bugs as your users no longer have a mechanism to communicate the bugs to you other than (1) paying you money or (2) posting to your forum, where you are normally a lurker.

Link to comment
Share on other sites

As you can probably guess from my posts in the companion thread, I agree with everything you say. The most annoying issue is not the slowdown and hangs, but the lack of a meaningful response from Dantz. Even if they've concluded that the problems are not caused by Retrospect per se, they should say so. (I doubt that this is the case, however.)

 

 

 

During the past couple of weeks, I've gone from a "casual" to a "frequent" forum poster, but I can think of many things that I'd rather be doing. :)

 

 

 

-waveman

Link to comment
Share on other sites

> 1. Have you solved them and, if so, how?

 

 

 

Not solved, still trying to find a configuration that works. I am currently connecting my VXA-1 drive through a Microtech scsi-firewire adaptor, rather than through an Adaptec 2930 card. Much too early to tell for sure, but I had a good run with almost 48 hours continuous network backup.

 

 

 

Unfortunately one client then stalled the backup for the next 30 hours. Unattended backups are hell at the moment!

Link to comment
Share on other sites

Dantz has been collecting information to find a pattern in these problems. In your case, waveman, a 25-pin 2906 card is not going to support a 68- or 50- pin device, especially under OS X.

 

 

 

I'm sorry that this information wasn't posted sooner; it would have been if the forum had not been down for maintenance.

 

 

 

Thanks,

 

 

 

Irena Solomon

 

Dantz Tech Support

Link to comment
Share on other sites

Here's a Knowledgebase entry that describes the pauses during the backup:

 

 

 

Retrospect for Macintosh 5.0 has a change in the built-in driver for the Ecrix VXA-1 drive. After working with Ecrix, Dantz implemented this to ensure proper data integrity after a power loss.

 

 

 

Retrospect issues a read command after the drive writes a file mark to the tape (apporiximately every 200 MB). When this happens, a brief pause or slowdown may be noticeable. This is the expected behavior and is normal during the copy phase of a backup operation.

 

 

 

We continue to work with Exabyte for a long-term solution that includes

 

improved performance as well as reliability.

 

 

 

Irena Solomon

 

Danzt Tech Support

 

 

Link to comment
Share on other sites

Irena,

 

 

 

Thanks for the post, but I'm confused.

 

 

 

I've been in touch with Exabyte's tech support, who are aware of the details of my setup, and they've not said anything about an incompatibility.

 

 

 

I'll certainly buy a new SCSI card if need be, but I want to make sure it'll work first.

 

 

 

Sincerely,

 

 

 

-waveman

 

 

 

Edit: BTW, I've been doing incremental backups of about 600-700MB without any problem, so I feel that I have some breathing room until my next full backup.

Link to comment
Share on other sites

  • 2 weeks later...

I had problems, and they are now solved.

 

I was using a SCSI VXA-1 and Adaptec 2930U (latest versions of firmware, BIOS, driver, etc.)

 

and seeing backups hang after around 1GB (this is under X/RD5.0, latest of each).

 

 

 

Everything appeared to be working fine with the same setup under 9.0.2/RD4.3,

 

but upon further investigation found that it was not: my backups would not verify.

 

(I had been running with verify off for ~1yr.)

 

I was seeing ~200 verify errors in a 20GB dump. Ecrix replaced the drive, thankfully

 

with a Firewire version. All works fine now: both 9.0.2/4.3 and X/5.0. 5.0 has the annoying

 

pause behavior, but at least it completes and verified.

 

It'd be great if the pauses were eliminated, but I can live with this for now.

 

 

 

Lessons learned:

 

1. If anyone ever writes a decent backup program for MacOS I will drop Retrospect like

 

a hot potato. A working UNIX dump/restore would be a godsend.

 

2. I'm never going to get a SCSI device for a Mac again. Firewire is the way to go.

 

 

 

PS Waveman, your persistence in trying to get answers to these problems is much

 

appreciated, at least by me.

 

 

 

 

 

 

Link to comment
Share on other sites

As I have posted, one of my shop's backups are running on VXAs, albeit at miserable rates.

 

 

 

But I have noticed another pattern that is beginning to unfold. Every few days I see a tape ejected from the VXA with a Retro(gress) dialog box saying that the drive isn't responding, but please put the same tape back in. At first this gave me the willies, because in the past Retro would make me put in a new tape. But the last time I saw this problem was several years ago when they were using DATs; I hadn't seen this before on a VXA drive. I got the courage up to shove the tape back in, and the backup apparently completed fine, without restarting Retro. I did go back and do some test restores, and as far as I could tell, Retro got everything on the tape and didn't overwrite the previous backup.

 

 

 

I went back through the logs, and sure enough this started happening exactly after switching over from Preview to Release. Nothing like this happened on Preview, or under 4.3. Of course I've made sure that the drive is cleaned on schedule, and there seems to be no correlation to when the last cleaning happened.

 

 

 

I don't normally feel I should have to babysit backups, but over the last few days I have been watching twice when this error occurred. Once was during one of those beloved backspaces, and the other happened when a client backup started writing, after another completed. Actually they may be completely related, because Release now seems to do the backspace after finishing each client; on 4.3 and Preview it did not and the next client backup started up a LOT faster. So yet another failure mode, all in the name of power failure integrity.

 

 

 

I'm feeling very betrayed. My Mac OS X Server transitions have been very rocky, partly because of the lack of a backup solution for so long. Dantz kept pointing the finger at Apple as the rationale why they couldn't release Retro for X. Finally Preview came, and although it was buggy, it did give me enough hope to tell my clients to hang in there.

 

 

 

Now we get this mess. I would be very understanding if there was some logic or bug behind this, but that doesn't seem to be the case. I can't ethically recommend my clients upgrade to Release now, so I've had to do some ugly hacks to keep their systems backing up at the performance level they need. I hate to admit it, but I may have to seriously contemplate dropping Retro, and therefore Macs because of this debacle.

Link to comment
Share on other sites

Irena,

 

 

 

I would like to know if I pay for a tech support incident, will it guarantee that I will be able to get a working backup with my VXA drive. I've posted all the configuration information that I can possibly post, I've offered the use of my Adapter card and I'll offer my VXA drive as well. I can do valid backups with Retro 5 on the OS 9 side, but with OS X--I'm stuck. Of course I only have to boot to OS 9 now to do backups and that is a real pain. In fact, I'm only getting backups done about once a week that way and I'm afraid of losing 6 days of work. What am I to do? Can Dantz provide me with a working solution? On my behalf, can they talk Exabyte into replacing my drive with a firewire version since that appears to be working for some people?

 

 

 

Thanks,

 

David

Link to comment
Share on other sites

  • 3 weeks later...

Greetings,

 

 

 

New Retrospect Driver Updates for current Mac and Windows versions of Retrospect are now available for download from the Dantz website.

 

 

 

Mac RDU 2.7 is recommended for all Retrospect 5.0 users. In addition to new drivers for a handful of previously supported devices, RDU 2.7 adds support for the following drives:

 

 

 

Benchmark VS640

 

Exabyte VXA AutoPak 1x10

 

EZQuest Boa CDR/w 24X10X40 FireWire Ext

 

HP SureStore Tape Autoloader 1/8 Series

 

VST SmartDisk FWCDRW-V8

 

 

 

It also includes improved handling of the following drives:

 

 

 

Exabyte/Ecrix VXA tape drives

 

Que! Firewire 8x4x32x CD drive (prevents the assert at cdr.c-6249)

 

 

 

The RDU may be downloaded from:

 

 

 

http://www.dantz.com/index.php3?SCREEN=rdu

 

 

 

Thanks,

 

 

 

Irena Solomon

 

Dantz Tech Support

Link to comment
Share on other sites

miw,

 

 

 

Thanks for the vote of confidence! :)

 

 

 

I haven't posted here in a while because my incremental backups, which are typically < 1GB, have been working fine. However, I am anxious to try out the new driver update to see if it makes any difference on my next full backup.

 

 

 

-waveman

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...