A Bug about the JVM Attach mechanism

[Peter Levart]

On 6/22/19 3:07 AM, David Holmes wrote:
>>>> So the solution could be for attach thread (if it is already 
>>>> started) to update mtime or ctime of the .java_pid<pid> socket file 
>>>> periodically so cleanup job would leave it alone.
>>>>
>>>> What do you think?
>>>
>>> I'm not keen on having the attach listener thread periodically 
>>> wakeup just to do this.
>>
>> The required frequency would be very low. Once a day would be more 
>> than enough. cron jobs are usually set-up to remove files that have 
>> not been touched for several days.
>
> I would have thought it more prudent to remove files for which the 
> owning process no longer exists. :(

I don't know if there is a way to reliably associate a file with an 
"owning" process. Sure /tmp/.java_pid<pid> has the process ID encoded in 
its name, but that is java specific and a generic cron job knows nothing 
about that pattern by default. If the process still has the file open, 
there is a way to detect that and "tmpwatch" for example has an option 
--fuser that skips open files. There are other tools for managing 
temporary files that can't do that (systemd-tmpfiles for example) and 
only use file's timestamps to determine if the file is still "alive" or 
not...

Regards, Peter

>
> Cheers,
> David