RFR: 7164841: Improvements to the GC log file rotation

Tao Mao tao.mao at oracle.com
Wed Aug 28 10:45:19 PDT 2013


Yumin,

To be honest, I hate to add any more flags too. I'm with you :-)

But, the fact is sometimes we have to. The way you proposed to parse the 
gc log file name is not that intuitive so we set one more block on the 
learning curve, whereas users seem to get used to set a flag. That's why 
I prefer just adding one more flag. At this point, you need to make your 
own call since you have seen all these arguments in this thread.

My 2¢.

Tao


On 8/28/13 10:34 AM, Yumin Qi wrote:
> I really hate to add more flags, though it does not hurt with one more 
> since we already have tons of them.
> I think if the pattern exists in file name, we can just replace them 
> with pid or timestamp so can save a flag:-)
>
> Thanks
> Yumin
>
> On 8/28/2013 10:30 AM, Tao Mao wrote:
>> How about a flag AppendDataAndTimeToGCLogFileName (and default is 
>> false) or the like?
>>
>> Thanks.
>> Tao
>>
>> On 8/28/13 9:59 AM, Yumin Qi wrote:
>>> Not sure customer really wants to have like
>>>
>>> -Xloggc:app.log[.%p][.%t]   (whatever the position of %p and %t)
>>>
>>> at case rotate gc log or not.
>>> I can add parsing for name pattern with %p and %t, but that really 
>>> needs to have an apprehensive consideration of such change.
>>>
>>> Any other opinion?
>>>
>>> Yumin
>>>
>>> On 8/28/2013 8:30 AM, STIRLING, SCOTT wrote:
>>>> Disk overruns can be limited with dedicated log partitions.
>>>>
>>>> Logs can be pre-parsed with command line text tools into separate 
>>>> runs, which everyone already has to do for stack traces in system 
>>>> out/err logs.
>>>>
>>>> IBM logs work great in PMAT, which also keeps pace with 
>>>> ever-evolving parsing nightmare of detailed Hotspot GC logging :-)
>>>>
>>>> Scott
>>>>
>>>> On Aug 28, 2013, at 11:13 AM, "Kirk Pepperdine" <kirk at kodewerk.com> 
>>>> wrote:
>>>>
>>>>>
>>>>>> Append vs clobber: with or without rotation, -Xloggc always 
>>>>>> clobbers the last log. That's why people want a time stamp or pid 
>>>>>> in the gc file name, IMHO.
>>>>> My client just clobbered his log files yet... I'd still *not* want 
>>>>> the time stamp because this can potentially accidentally fill 
>>>>> production disks. Better that my client over wrote the logs than 
>>>>> filled his production disk because he didn't realize the 
>>>>> behaviour. And, many don't realize the behaviour, they just follow 
>>>>> the admin manual.
>>>>>
>>>>>> Alternatively, give an option to append when -Xloggc is enabled, 
>>>>>> vs default overwrite.
>>>>> Allowing an append will break all current analysis tooling.
>>>>>
>>>>> Regards,
>>>>> Kirk
>>>>>
>>>>>
>>>
>


More information about the hotspot-runtime-dev mailing list