Welcome Guest! To enable all features please Login or Register.

Notification

Icon
Error

Forum Jump  
massive problems with Logic/Network -can anyone advise?
Last post Fri, Mar 02 2007 by cm, 4 replies.
Options
Go to last post
Posted on Tue, Feb 27 2007 13:20
by Nigel Watson
Joined on Fri, Dec 20 2002, Cologne, Germany, Posts 365
Is anyone out there using Logic on a network together with other composers, with shared samples and audio on an external RAID system via File Server? If you are, can you please offer some advice on what does work, because at the moment our system does not!

We are a studio with 3 workspaces & 3 G5 Macs, making music for TV and media. Our samples and our music archive are on 2 separate RAID 5 configurations of about 2TB each. The samples are on one of the RAIDS, most of the music Archive on the other RAID. A G4 Dual operates as file server. Our RAID system hardware is custom, not Apple's X-serve.

This system served us well for a year or two, but now we are having extreme difficulties for the second time in several months. The system has actually had a heavier load because in the last few months we went from 2 Workspaces to 3, but it's hard to make a direct correlation between this and the current problems.

What's happening is that repeatedly the data on the RAID system becomes unavailable -samples get cut off, and in exs24 you can see the "couldn't get samples in time" error. The problem is probably more serious than just a bottleneck of data because when it occurs, files also become inaccessible via the Finder: in other words, we can navigate through the folder structure within the Finder, but the individual files themselves can't be previewed as is usually possible from the Finder; when you click on the Preview triangle (for an Audio file let's say) there is the message "fetching", and you wait....and wait. Generally a Restart of Logic, or a Fileserver Restart cures the problem but only for a very short time (sometimes minutes, sometimes hours) before it crops up again. Basically, communication between the Workspaces and the RAID system at file level is hung up or lost completely/

This problem caused our File Server computer to have Kernel Panics (last Autumn) and we thought there was a Data Registry problem on one of the RAID systems, so we re-initialized the RAID hoping that would cure it. Things went fine for a few months, but last week the File Server had a KP again and according to my computer Tech the problem is similar to the one we had before.

We have multiple deadlines every week, and I originally chose this system for its capacity and reliability as our organization has expanded. As a temporary workaround we can put essential data on the individual computers, but that is of limited use (limited capacity) and creates major problems knowing who has the most recent versions of a project. At the weekend we will try out a different File Server configuration - a G5 and a more powerful Switcher. What I don't know is whether the problems are hardware-based or actually a result of using Logic on a Network, which perhaps it was never intended for.

Any feedback/advice/suggestions would be much appreciated. We're all on Logic 7.2.3 under OS 10.4.7.

Thanks very much

Nigel
Posted on Tue, Feb 27 2007 19:21
by cm
Joined on Fri, Dec 20 2002, vienna, Posts 9101
nigel, your description is kind of vague regarding the technical aspects ... ahve you considered the raid controller itself to be faulty ... or the cache memory of the controller ... or some cables to the drives.

also does the G4 (server) have gigibit as well as the clients and is the used (gigabit-)switch healthy and reliably working?
if it's a 100 Mbit network 3 clients might be too much, whereas 2 would work fine ...

kernel-panic ... you might have a corrupted TCP/IP stack on your server - see if you can upgrade to 10.4.8 to solve such an issue (create an image of the system first!)

does the system log hand out any detailed information to guess where to start research?

what about keeping the projects on a server-share and at least some sample data locally (name the location to the same on every client to keep filepaths consistent)
and remember: a CRAY is the only computer that runs an endless loop in just four hours ...
Posted on Thu, Mar 01 2007 23:54
by Nigel Watson
Joined on Fri, Dec 20 2002, Cologne, Germany, Posts 365
cm,

I appreciate your post, thank you. I'll pass your suggestions on to my Tech.

One thing we didn't know is that a RAID 5 array-which is what we have- drops rapidly in performance when over 80% full. We've reached that level so obviously that is the first place to start tackling the problem. In that respect, increasing capacity is clearly neccessary, and it could be a good approach to take advantage of newer, bigger drives and keep some data locally on the individual workstations as you suggest.

The crash log on the File Server reports "trying to write data to corrupt node". That doesn't give direct clues as to where the problem lies, only perhaps to its nature.

We do have Gigabit, and we will do some testing at the weekend to see if a better switcher improves the performance. The file server is actually on 10.4.8, just the workstations are not yet.

Somehow it's been a lucky week and after a very trying start (when I wrote this original post), we had no more problems and were able to work normally. I'm grateful for that. Those small hours of the morning when the deadline is creeping up and the system is not working properly can be quite a trial.

I'll report when we've done some testing-thanks for any further input

Nigel
Posted on Fri, Mar 02 2007 05:57
by cm
Joined on Fri, Dec 20 2002, vienna, Posts 9101
Quote:
drops rapidly in performance when over 80% full
good point - this can happen in certain cases if the inernal organization of data becomes too fragmented. most stand-alone raid5 solutions have a feature called *re-organization* or similar (the de-fragmentation utilities of the operating system are of no use in such a case because a kind of filesystem virtualization happens on raid5 devices)
christian
and remember: a CRAY is the only computer that runs an endless loop in just four hours ...
You cannot post new threads in this forum.
You cannot reply to threads in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.