Page 2 of 2
Re: Sab 0.5 - 0.5.2rc1 causing OS X 10.6 Crahes!
Posted: June 29th, 2010, 5:15 pm
by x0x
Yeah I will find the cause for this heh.
Re: Sab 0.5 - 0.5.2rc1 causing OS X 10.6 Crahes!
Posted: July 2nd, 2010, 3:54 pm
by x0x
Update from my side.
I fully wiped my HDD again and reinstalled a fresh copy of Mac OS X. Installed SAB 0.5.3, added a few servers and had it download with a total of 30 connections. After not so long, the Mac completely crashed like before.
So the problem is not related to iStat and/or other software and more and more seems related to an issue with many connections at the same time. Same as has been reported with torrent client Transmission. Problem caused by the network driver?
Question is, should I take this up to Apple or is there something the SAB developers can do?
Re: Sab 0.5 - 0.5.2rc1 causing OS X 10.6 Crahes!
Posted: July 3rd, 2010, 3:30 am
by x0x
Another update. I added someone's Giganews account with 50 connections. The Mac completely crashed within 2 minutes. This is a serious problem but I think it's the combination of a certain netwerk driver in OS X and SAB. Possibly only with 0.5?
Re: Sab 0.5 - 0.5.2rc1 causing OS X 10.6 Crahes!
Posted: July 3rd, 2010, 4:59 am
by shypike
If Transmission shows the same problem, then it does point to an issue in OSX.
So far the work-around is to lower the number of connections.
Most servers don't need many anyway, despite the "more connections" hype.
When our OSX expert returns from holiday, he has his work cut out
Re: Sab 0.5 - 0.5.2rc1 causing OS X 10.6 Crahes!
Posted: July 3rd, 2010, 5:02 am
by x0x
Yes. I have tested this on a MacBook Pro and that does not seem to have the problem. Maybe it's the NIC from the new Mac Mini 2010 or it's driver software.
Re: Sab 0.5 - 0.5.2rc1 causing OS X 10.6 Crahes!
Posted: July 12th, 2010, 10:19 am
by primalyeti
So, for the past couple weeks I thought I had solved the issue.
The way I went about it? I installed Sab on my virtual copy of Windows 7 and just ran it in Parallels.
This worked for several weeks (maybe even over a month), with small and large files, long and short queues, with all the exact same settings as Sab on OS X (with the exception of folders).
Now today, all was going fine and then my net died, I tried to go in to my Win7 vm and it had crashed...
So it would seem this issue is not OS dependant.
Re: Sab 0.5 - 0.5.2rc1 causing OS X 10.6 Crahes!
Posted: July 12th, 2010, 12:38 pm
by rAf
Hi,
I'm back
I think your setups are hard to debug...
Crashes may be caused by :
- SMB mount disks
- Numbers of servers
- Numbers of connexions
- High bandwidth / Slow disk writes
- ...
We should go by steps :
- Reinstall SABnzbd with one server, a few connections (10 or 20) and local folders for incomplete and complete.
If everything works fine, we'll add more server then more connexions and then smb mounts...
It can help us to find us what is going on...
Your thoughts ?
Re: Sab 0.5 - 0.5.2rc1 causing OS X 10.6 Crahes!
Posted: September 22nd, 2010, 2:30 pm
by c0ckster
I've been having the same issues - Mac Mini 2.0ghz SL 10.6.4. I thought it might be hardware related so I reset PRAM and SMC but it still persisted. I'm using Astraweb and I dropped my connections to 20 - so far so good.
I'll check back after a while to report if the fix holds.
Re: Sab 0.5 - 0.5.2rc1 causing OS X 10.6 Crahes!
Posted: October 22nd, 2010, 5:11 pm
by how_gauche
Having this issue also. Betcha any money the kernel is running out of network buffers; running "netstat -m" on my system:
Code: Select all
$ netstat -m
5309/7473 mbufs in use:
5176 mbufs allocated to data
5 mbufs allocated to socket names and addresses
128 mbufs allocated to packet tags
2164 mbufs allocated to caches
6143/10922 mbuf 2KB clusters in use
2/16 mbuf 4KB clusters in use
0/8 mbuf 16KB clusters in use
22037 KB allocated to network (52.9% in use)
0 requests for memory denied
169 requests for memory delayed
103 calls to drain routines
This is *after* I lowered my # of connections value from 50 to 20, which solved the issue. Note the "requests for memory delayed" -- I think the kernel is running out of network buffers, which is why all of the other network activity on the box grinds to a halt. Once apps start getting their requests for network resources denied, the box gets into a logjam situation from which there is no alternative but to reboot.
Try increasing your mbufs setting and rebooting to see if the problem goes away:
Code: Select all
$ sudo nvram boot-args="ncl=131072 initmcl=131072 nbuf=131072"