Crashes always preceded by kernel[0]: zalloc did gc

[sman591]

I've gone and done the same, installed FakeSMC from MB 5.3.1. I'll keep an eye on kernel_task and see if it fixes the issue

 

[ppohio]

Whooooo - feeling pretty good about this right now, as my kernel_task did something that seemed unthinkable last week: it actually went DOWN. Doing some RAM-heavy tasks throughout the night (transcoding a day's worth of footage), will report back in the morning.

 

[wonkywonky]

@ppohio Good to hear!

In other news, I checked out kozlek's FakeSMC repo. Here's hoping the leak(s) that killed us have been fixed

 

[sman591]

Same here, kernel_task has been staying right around 1GB since I switched FakeSMC last night.

@wonkywonky fantastic! Looks like that was the exact problem we were getting. Hopefully it'll be out soon...

 

[sman591]

Almost certain that it's solved now. Looking at the iStat Server graphs compared to what they've been like for the past few weeks, the wired memory is actually staying flat!

 

[wonkywonky]

Excellent. I'm glad I posted here - after spending 3 weeks banging my head against the wall wondering what I did to break my machines and finding nothing useful in Google, I was so sure at one point it was just me, and was this close to wiping everything and doing a fresh install (which would have ended up solving nothing, as it turns out).

 

[ppohio]

Has anybody tried the updated Multibeast 5.5? Looks like there's changes to the FakeSMC stuff in there. I am on deadline until the 21st so I haven't taken the plunge yet...

 

[wonkywonky]

Has anybody tried the updated Multibeast 5.5? Looks like there's changes to the FakeSMC stuff in there. I am on deadline until the 21st so I haven't taken the plunge yet...

Yep, works great.

 

[sman591]

Has anybody tried the updated Multibeast 5.5? Looks like there's changes to the FakeSMC stuff in there. I am on deadline until the 21st so I haven't taken the plunge yet...

Been using Mavericks with Multibeast 6.0 for about 24 hours now, no issues.