On 03/09/10 13:44, John Baldwin wrote:
> On Monday 08 March 2010 5:59:29 pm vol...@vwsoft.com wrote:
>> On 03/08/10 12:16, Giulio Ferro wrote:
>>> Freebsd 8 stable amd64
>>>
>>> It mounts different file systems by NFS (with locking) on a
>>> data server directly connected (gigabit) to the server
>>>
>>> Apache running in a several jails on those nfs folders.
>>>
>>> Now and then I get huge slow-down. When I look in the logs
>>> I get thousand of lines like these:
>>> Mar 5 11:50:52 virt2 kernel: vm_fault: pager read error, pid 46487
> (httpd)
>>> Mar 5 11:50:52 virt2 kernel: pid 46487 (httpd), uid 80: exited on
>>> signal 11
>>>
>>>
>>> What should I do?
>> Giulio,
>>
>> it seems this is anyhow not related to network (nfs) operations. It's
>> looking like a problem in the VM. I think it makes sense to have a look
>> at the httpd.core file if the binary has been linked with debugging
>> symbols turned on. Also I think at first, it may not hurt to look at
>> vmstat -m output.
>>
>> You may want to change ${subject} and post to stable@ to drive more
>> attention to your problem.
>
> That's not quite true. If you take a page fault on a mmap'd file backed by
> NFS (e.g. an executable or shared library) and an NFS READ RPC to satisfy the
> page fault fails, then you could get this error.
>
John,
thank you for pointing that out. I've forgotten the mmap'ing of files
over nfs as a possible source of that problem.
With 8-stable I'm seeing mbufs leaking with nfs operation. It may or may
not be related to Giulio's problem.
Volker
_______________________________________________
freebsd-hackers@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-hackers
To unsubscribe, send any mail to "freebsd-hackers-unsubscr...@freebsd.org"
