Plex Server Cache over 310GB

[sa2000]

@croneter

My guess:

1. PMS scans metadata of this “corrupt” video
2. PMS tells anyone listening on websocket that scan is finished (and element thus potentially changed)
3. PKC wants to grab new metadata for the corrupt video and requests metadata from the PMS (what you see in the PMS log)
4. PMS sees that the item is corrupt and concludes that the video needs to be scanned again -> start again at 1.
5. Continue for eternity

Would have to be fixed on the PMS side, then.

Thanks for your input. I will have a look at what websocket events get generated

[kantlivelong]

Logs sent via PM and opened PKC ticket.

[sa2000]

@croneter

PKC won’t ask twice if data is missing.

My guess:

1. PMS scans metadata of this “corrupt” video
2. PMS tells anyone listening on websocket that scan is finished (and element thus potentially changed)
3. PKC wants to grab new metadata for the corrupt video and requests metadata from the PMS (what you see in the PMS log)
4. PMS sees that the item is corrupt and concludes that the video needs to be scanned again -> start again at 1.
5. Continue for eternity

Would have to be fixed on the PMS side, then.

I have been looking into the websocket notification and also how Plex Web handles them.

I would like to suggest a small change to the requests PlexKodiConnect makes

The request in question that has been causing the refresh loop was
GET /library/metadata/xxxxx?includeRelated=0&includeExtras=1&includeReviews=1&checkFiles=0

For step 3 in your My guess list, i would like you to add this parameter to the request skipRefresh=1

So if you are making a request for the new metadata because of a websocket notification telling you that the item was scanned - then please add the parameter that stops a further refresh being made.

I have tested it and it does not result in another on-the-fly scan. The initial request is fine and does not need the skipRefresh=1 but those requests made because a scan/refresh has just completed ought to set the parameter to make sure we do not refresh again (for bad media)

Hope it is not too difficult a change

[Adenn]

Do you have “Generate Thumbnail Previews” turn on in your settings > server > library ?

My cache and such was VERY small until I turned that on. My cache then skyrocketed to over 250 GB.

If you don’t use that feature and it is turned on, try turning it off and see what happens to your disk usage.

I quite like that feature so I take the drive space hit and leave it on.

[croneter]

Ouh that’s great, that skipRefresh=1 will probably solve another long-standing PKC issue. Many thanks!! Fix will be included in next stable and beta version of PlexKodiConnect

[kantlivelong]

So it looks like the change @croneter made has definitely reduced the duplicates but they are not eliminated. Not sure if this is something Plex would still look into or not.

If not, is there any kind of database that keeps track of these cached images that can be queried?

[sa2000]

@kantlivelong

So it looks like the change @croneter made has definitely reduced the duplicates but they are not eliminated. Not sure if this is something Plex would still look into or not.

I believe it would only arise if different elements end up using an identical image file - perhaps even obtained from a different source. It should not have any big impact and there would be no spiralling escalation of disc usage

@kantlivelong

If not, is there any kind of database that keeps track of these cached images that can be queried?

There is no public information on how Plex Media Server manages its own internal databases and file structures and how and what is done today may change