Still suffering from insane save times

[Prime Mover]

I think I posted about this a month or so back, but I can't find the old post to amend, sorry for the new one.

I've still been suffering under these 3-5 minute epic save freezes. It's not all saves, some of them just take moments. Then the next will be 5 minutes. It's one of the reasons I've been posting so much on the forum! I have to sit for 5 minutes while it beach balls, and I have nothing better to do!

Anyway, there doesn't seem to be any pattern to this problem. New project files, old project files, big ones, small ones, ones that have just had their undolist pruned, others that have been building up for weeks... doesn't make a difference, they all suffer from it. I followed my activity monitor last night and was pretty interested in what I found. The disc IO read and write skyrockets throughout, but the project file in question yesterday was only 15MB. I was getting 10Mb IO for a good 4 minutes... what's all that data? And why is it reading so much? One would think writing a file would be largely a data write process, but the read rate was just about as high as the write rate! CPU and RAM are pretty moderate, so it's not like it's doing a huge amount of processing. I did a search on local files to see if anything external was being written, but no, the only thing saved at that time was the project file itself... analysis files, undo list, soundbites, nothing got modified during that period, even system and cache files. So whatever is going on, it's the project file that's getting written.

Time machine was connected but not updating, there doesn't seem to be any correlation there. Network activity was low, so it wasn't trying to do anything over the network or on other drives either! I believe I'm encountered this when completely out of my network or Time Machine, though I could shutoff the WiFi just in case. But I don't think that's it.

But the whole fact that it SOMETIMES does this and sometimes doesn't is just bizarre. I have it autosave every 10 minutes, and I probably get this one out of every 2 or 3 times. Sometimes my manual saves do the same thing, but at about the same frequency.

I made 3 major changes within about a month, about 4 months ago: upgraded to DP8, upgraded from SL to ML, and went 64bit with jBridgeX. I can't be sure which one was the culprit. Obviously, many people are using the same configuration and not having this problem, so it's not one of those things alone.

And I've already tried trashing prefs and fixing privileges, so that's not it either.

Thoughts?

[MIDI Life Crisis]

I'd run disk utility and verify all drives. It could be your file directory is starting to go or even a bad sector.

[Prime Mover]

Yeah, I dunno. I ran Disk Utility after I started having this problem, and it all checked out. It wouldn't surprise me if my drive is having problems, I'm working it REALLY HARD, it gets very hot... I sometimes throw it outside at -30F to get it to cool down in the middle of a recording session. But Disk Utility tells me everything is fine.

But weird thing is, I've never had this problem with any other programs, and all DP projects (which are probably scattered throughout the platter), so I'm not completely convinced it's a physical disk problem. I'm pretty tempted to blame DP8 for doing something weird, maybe a conflict that's causing it to keep writing to the harddrive over and over again. What I'm seeing from activity monitor is that it's writing A LOT of data, so it just must be writing things over and over again.

I just realized that I also replaced the RAM around that time... maybe I should check for a bad DIMM. That could cause some difficulty for file saving!

[cuttime]

Try flushing Undo History?

[Dan Worley]

I just realized that I also replaced the RAM around that time... maybe I should check for a bad DIMM. That could cause some difficulty for file saving!

That's where I found the problem to be when this happened to me sometime ago. I had to take dimms out one by one until I found the bad one. Testing them didn't show any problems.

[Prime Mover]

Hmmm, maybe I'll give it a shot. I'm running memtest right now, and so far, everything checks out.

AHAH, I think I may have a possible explanation, or maybe a fix. All the files I've been using are old DP7 files. I didn't think they were, because many of them had been duplicated and trimmed, and I had been thinking of them as new files, but I think they all stem from the same DP7 project file. I know this often causes problems. I'll load it into a fresh DP8 file and see how it goes.

EDIT: Yeehhaaa!!! I think that fixed it (crosses fingers). I saved about 8 times after I had loaded the file over, and all the saves were fine. One was about 15 seconds, and I got worried, but it finally stopped. Everything else was between 1-8 seconds, which was what it used to be. Each time, I did a select-delete of a blank area, which adds an element to the undo list and fools DP into thinking it's been edited.

Unfortunately, the loading over was a bit of a hassle. With the Comp region bug, you can't simply load over the project normally. I used the load command to load over EVERYTHING but the chunks (actually, I loaded 2 other chunks that didn't use the comp tool). Then I went back to the original project, went to the chunks window and copied the chunk manually, and pasted it into the new project. This preserves comp regions, but unfortunately, doesn't reconnect IO, everything is italic. Then it's just a question of going through and manually setting the IO. Simple for basic track IO (option-A), kind of a pain for send IO, and really scary for sidechain IO since some plugins completely forget their bus setting. Thankfully, I wasn't using many sidechains.

I really wish DP had a command that would go down through and reconnect all bundles. I mean, it already knows the correct name for the bundle, it should be fairly simple to tell it to just "reconnect with the bundle of the same name".

[Prime Mover]

Spoke too soon, I got a minute one just now. We'll see...