Page MenuHomePhabricator

Wikimedia Italia's is huge (2G database and 4G of files) - reduce footprint
Closed, ResolvedPublic


The user gvf reported that ("la nostra wikina") is somehow huge. It currently has 4G of files and a 2G database.

To reduce the disk space usage, purge the page and thumbnail cache.

Event Timeline

It actually consists in 4G of files and 2G database.

Why would that be a problem?

It actually consists in 4G of files and 2G database.

Why would that be a problem?

We should ask gvf asd

Nemo_bis updated the task description. (Show Details)
Nemo_bis closed this task as Resolved.EditedSat, Jun 12, 9:48 AM
Nemo_bis claimed this task.

We should ask gvf asd

Ok, if gvf says 4 GB of files is too much we need to act on that.

As for the actions proposed in the task description, which I suppose are your own doing, they're not correct. Most of the size on disk is for the page and thumbnail cache, which have nothing to do with the size of local uploads. The disk cache is/was the cheapest way to reduce our impact on CPU and RAM (and it's not supposed to be included in backups either, as there are .nobackup files in those directories). This caching can be reassessed now that we're using PHP 7 and so on (after MediaWiki is upgraded), and if gvf prefers us to use more RAM and less disk that's one criterion to use, but for now the process is simply to purge those directories every now and then; the easiest way is for someone with shell access to issue a relevant find command, but we can also do it through SFTP.

The HTML cache compresses by a factor of 30 when 7z'ed:

2021-06-12 11:51:27          300077599      9385336  15492 files, 844 folders

Also, I have no idea why T279145 was added as subtask, but it's clearly not blocking the fix of this task so I'm removing it. If you created that task for the sole purpose of supposedly helping with this task, please close it as invalid.

TL;DR: Exterminate:

$ du -shc *
317M    cache
1,7G    thumb
2,0G    totale