[OpenAFS-port-darwin] Re: Many crashes on OS X 10.3

[Shantonu Sen]

You don't need to be actively using AFS for it to cause panics and 
system instability. In this case, I would not be shocked if AFS did not 
mark some vnode's state properly, and under heavy vnode load (like a 
backup, which may traverse thousands of files), the vnode tried to get 
recycled.

You might try tweaking your configuration to see if you can affect the 
frequency of panics. For instance:
[shantonu@ssen-albook]$ sysctl kern.maxvnodes
kern.maxvnodes = 9216
[shantonu@ssen-albook]$ sudo sysctl -w kern.maxvnodes=20000
Password:
kern.maxvnodes: 9216 -> 20000

Warning: this may increase the memory requirements of your system. You 
can probably also disable AFS from starting at boot to try to isolate 
the issue. I find it difficult to believe backing up to a firewire 
drive causes panics in isolation.

Shantonu

On Feb 6, 2004, at 12:02 PM, Steve Lidie wrote:

>
> On Feb 6, 2004, at 2:55 PM, Massimo Marino wrote:
>
>> On Feb 6, 2004, at 8:43 PM, Steve Lidie wrote:
>>
>>>
>>> On Feb 6, 2004, at 1:40 PM, Massimo Marino wrote:
>>>
>>>> First: I am not flaming but it might be important to know for 
>>>> people in charge that it is not AFS per se: I use
>>>
>>> Until this thread, I've been leaning towards Retrospect: I've 
>>> correlated the "vnreclaim" panics with unattended backups overnight. 
>>>  What I've discovered is that, backing up to an external FireWire 
>>> drive, I could reproduce a panic if the disk was full- i.e. no room 
>>> for Retrospect's catalog information. I've forwarded the panic logs 
>>> and my "diary" to both Apple and Dantz, but, as you might guess, 
>>> they both told me to go away.   Then again, perhaps both afs and 
>>> Retrospect are tweaking a Panther end case ...
>>
>> I also have and use Retrospect Express but am not backing up to AFS 
>> volumes - if that is what you imply - nor
>
>
> No - afs is not even involved - I'm just backing up the internal disk 
> to an external FireWirwe. I was just agreeing with you that it might 
> not be AFS.
>
>> external drives: my backups go to an online server. If and when there 
>> is no space the bck simply fails. I tuned that down re-cycling the 
>> bck after so many days automatically so not to run out of quota.
>>
>
> Oh, I can typically avoid the out-of-space condition.... but when I do 
> not I get the vnrelcaim kernel panic, and can reproduce it doing a 
> simple RS backup to an external drive.
>
>
> _______________________________________________
> port-darwin mailing list
> port-darwin@openafs.org
> https://lists.openafs.org/mailman/listinfo/port-darwin