Re: [pbs-devel] [PATCH proxmox-backup v2 1/3] docs: centralise and update garbage collection description

["Fabian Grünbichler" <f.gruenbichler@proxmox.com>]

On April 10, 2024 3:38 pm, Hannes Dürr wrote:
> 
> On 4/8/24 11:20, Fabian Grünbichler wrote:
>> On April 5, 2024 3:05 pm, Hannes Duerr wrote:
>>> +Chunks accessed after the cut-off time are marked as *Pending removals*
>>> +by the GC as it cannot be certain whether they are still needed.
>> this is rather incomplete and a bit hard to parse as well. I'd replace
>> "accessed after" with "with an atime after".
>>
>> pending is actually:
>> - chunks with atime between the cut-off and the oldest writer (if one
>>    exists)
> At this point i am slightly confused as we defined earlier:
> the cut-off is the start of oldest backup writer* (if one exists)

the cut off is the minimum of (oldest worker start, now-24h) (minus 5
minutes). there can be an oldest worker that was started after the cut
off timestamp, in which case GC might find pending chunks (see below).

> 
> Which would lead to the following:
> 
> - chunks with atime between the cut-off (which is the start of the 
> oldest existing writer) and the oldest writer (if one exists)
> 
> which does not make any sense, where is my mistake ?

if the cut off is determined by a worker started more than 24h before
the start of the GC, then there cannot be any pending chunks - because
all the chunks which might be considered pending (otherwise) could have
been written by that worker, we can't tell.

pending chunks can only happen if there is
- no backup writer
- the olders writer was started less than 24h before the GC

then any chunks written in the time frame between cut-off and writer
start (or GC start, if no writer exists), which are not yet referenced
by any snapshot/index, are considered pending. because those chunks are
neither referenced nor can they have been written by any still going
writer, so they are most likely "garbage".