On 02/23/2011 10:00 AM, Fabio M. Di Nitto wrote:
> On 2/23/2011 5:55 PM, Steven Dake wrote:
>> On 02/23/2011 06:28 AM, Fabio M. Di Nitto wrote:
>>> On 2/23/2011 2:17 PM, Russell Bryant wrote:
>>>> On Wed, Feb 23, 2011 at 1:53 AM, Fabio M. Di Nitto <[email protected]> 
>>>> wrote:
>>>>>> I hope the change from:
>>>>>> read (fd, &memb_ring_id->seq, sizeof (unsigned long long));
>>>>>> to
>>>>>> read (fd, &memb_ring_id->seq, sizeof (uint64_t));
>>>>>>
>>>>>> won't cause any problems.
>>>>>
>>>>> On i686 i noticed the file being 8 bytes (unsigned long long)... I
>>>>> wonder if you shutdown corosync, update packages with the fix, then
>>>>> restart.. is it going to read garbage from the file?
>>>>
>>>> but sizeof (uint64_t) is also 8 bytes.
>>>
>>> ok... i clearly need more coffee :))) ok either x86_64 or i686 had a 4
>>> bytes file and the other 8... one of them is going to be affected by
>>> switching to a different size.
>>>
>>
>> Are you serious?  It should be 8 bytes always!  Could you give more
>> details of your platform information (was it linux, which os version, etc)
> 
> They were 2 VMs RHEL6.0+z one i386 and one x86_64.
> 
> it´s entirely possible that the files were truncated somehow.. or that I
> do not remember properly.
> 
> Probably the same reason why x86_64 had a 0 bytes file.. go figure.
> those vms are long gone now and as long as you have tested it, I am OK
> with that. Don´t get my doubts in your way.
> 
> Fabio

If there is some wierd platform specific issue that was making the data
4 bytes on disk, the new code will convert it to 8 bytes on disk.  If
inside corosync it is actually 4 bytes (which unsigned long long
shouldn't be on rhel 6..:), (it is unsigned long long there in data
structures as well) We would have serious i386->x86_64 communication
problems which we don't see.
_______________________________________________
Openais mailing list
[email protected]
https://lists.linux-foundation.org/mailman/listinfo/openais

Reply via email to