View Single Post
Old 09-29-2009, 03:16 AM
Londoner Londoner is offline
Join Date: Feb 2009
Location: London
Posts: 18
Default 6086, 6097, 9128 errors solved

I post this so that others might learn from my mistakes. If you want to skip to the punch line, scroll to the bottom.

I'd been running Pro Tools LE on my PowerBook G4 happily for 5 years, with a 002 Rack, without any external disks. My sessions consisted of a lot of MIDI tracks, powering two old Emu sound modules which fed into one audio track.

Six months ago I installed Kontakt 3, and that too worked fine, perhaps because most of my MIDI tracks remained on the Emu hardware modules. I installed the maximum memory on the PowerBook, and upgraded to the latest versions of Pro Tools and OS X which my system could handle.

Then one of the Emu boxes died, and I had to make much more use of Kontakt. My latest session, as I continued adding to it, experienced more and more problems. I had four discrete sets of symptoms:

Symptoms 1 (the really important ones): Errors 6086, 6097, and 9128 at random times. Sometimes error 6097 appeared before a single note was played. Often (but not always), an error occurred just when a previously-unused Kontakt instrument started to play.

Symptoms 2: Activity Monitor showed sharp 95% spikes in CPU usage by system tasks (not user tasks), every 80-100 seconds. I don't know how long this had been going on; I only noticed it after starting to diagnose the first set of symptoms. It was not related to appearances of errors 6086 or 6097, and many 9128 errors occurred when total CPU usage was below 75%. This symptom was not limited to Pro Tools; any app which used FireWire (eg. iTunes and Finale) provoked it.

Symptoms 3: After the other two symptoms had been appearing for a few weeks, I became unable to open my session by double-clicking it. I had to first start Pro Tools, then open the session; otherwise the session start-up collapsed midway.

Symptoms 4: In /var/log/system.log after every shutdown was initiated, there was a message from SystemStarter: "Digidesign Mbox 2 Configuration Service (nnn) did not complete successfully". Sometimes it also reported: "/Library/StartupItems/Digidesign Mbox 2 - execution of Startup script failed".

Changes I made, and their effects:

Upgraded Kontakt from 3.0.2 to 3.5 which claims to use less CPU power. No visible improvement.

Told Kontakt to hold a couple of the most-used MIDI instruments' samples permanently in memory, to reduce disk I/O. No improvement.

Replaced the FireWire cable between the PowerBook's built-in port and the 002 Rack. No improvement.

Removed the entire folder "Digidesign Mbox 2" from /Library/StartupItems. This eliminated symptom 4.

PRAM reset by holding down Command+Option+P+R while restarting (causes Mac to rescan its config). Either this, or the StartupItems action above, eliminated symptom 3.

Changed HW buffer size and back. No improvement. Still had symptoms 1 and 2.

To test my built-in FireWire port as a potential cause of symptom 2, I tried a Pluscom 2-port 1394 FireWire Cardbus card with VIA VT6307 chips, but every time it was in the slot, the system told me to restart. I couldn't get the computer to function at all when the card was in the slot.

Bought an OWC Mercury Elite-AL Pro external disk, with 16MB cache and two FW400 ports. Attached it to my built-in FireWire port, then daisy-chained the 002R off the disk enclosure's second port. I cannot honestly say that it's a very quiet disk; it makes about as much noise as the PowerBook.

Since the OWC comes preformatted for Mac (in GUID format), I copied the session folder and all its contents to the OWC and opened the session. Mistake! Error 9131 caused me to search the DUC, finding many red herrings before finding a page which was not in the "9131" search results, and which said I could not use the GUID format for audio. So I reformatted the disk as APM. No more 9131 errors.

Despite the investment in the disk drive and the hours spent searching the DUC, I continued getting 6097, 6086, and 9128 errors. 6097 always appeared when I opened the session, before playing a note. 6086 seemed usually to happen when a previously idle Kontakt instrument came in.

Bought a Texas Instruments FireWire PCMCIA card (model EX-6600E from Exsys). Unlike the Pluscom card, this didn't prevent my Mac running.

Moved the 002R's cable onto the PCMCIA card. Symptom 2 disappeared! No more big CPU spikes So that problem was due, somehow, to the built-in FireWire port.

6086, 6097, 9128 errors persisted.

Moved the OWC disk cable to the PCMCIA card, so that both the disk and the 002R are on that FireWire bus, and nothing is on the questionable built-in bus.

6086, 6097, 9128 errors persisted.

Attached a second OWC external disk to the built-in FireWire 400 port, moved all Kontakt samples to it, and told the Kontakt RTAS to take samples only from that location. Unfortunately this second OWC disk is much noisier than the first. I have to put it on a thick foam pad to avoid doubling the noise in my studio. If one disk is that different from another, you wonder about their quality control.

6086, 6097, 9128 errors persisted, PLUS now error 9403 occurred when the session opened. "You are running out of CPU power. Elastic Audio processing could not keep up. Please ... disable Elastic Audio ..." but it was ALREADY disabled in the only audio track.

So -- I had followed all the advice from Digidesign and NI, I had spent a lot of money and time, my work environment was more cluttered and noisy, and my electricity bills will go up -- and I had worse problems than when I started.

It seems I'm on my own.

My next approach was to create an entirely new PT session and gradually build it up with components from the problem session. Without any MIDI tracks, got error 6097 as soon as I loaded the Kontakt instances. Loading one MIDI violin track caused a 9128 error. But an older session, with 27 MIDI tracks, works flawlessly -- maybe because that session uses no Kontakt string instruments?

Told Kontakt to keep most violin samples in memory. This played the single-track session for 3 minutes perfectly until encountering one violin Program Change that I hadn't held in memory.

Lesson learned: Kontakt's dynamic loading of samples (DFD) was causing the problem. I was loading (not necessarily using) more DFD instruments than my 6-year-old computer could handle.

I reduced the total number of Kontakt instruments from 62 to 33. (I had been keeping several articulations in each bank, such as forte-piano, sforzando and staccato articulations for the violins, when starting to compose, not knowing which articulations I would end up using. Where they have actually been used, I've now produced a similar effect by fine adjustments of velocity, volume and duration.)

- No more 6086 errors
- No more 6097 errors
- No more 9128 errors
- No more 9403 errors
- Lots of money spent
- Lots of time wasted

But at least I can now do my work. I hope this helps someone else.
Reply With Quote