⚠️ maintenance ⚠️
Some general maintenance at 12:00 CET which involves quick reboots on mountains.social and mountains.photos.
⚠️ maintenance ⚠️
Some general maintenance at 12:00 CET which involves quick reboots on mountains.social and mountains.photos.
⚠️ maintenance ⚠️
Some general maintenance at 12:00 CET which involves quick reboots on mountains.social and mountains.photos.
This is the graph of the size of the media bucket over the last 3 months. Clearly visible when the scraping started. The drop in size at the end is me cleaning up after 3 instead of 4 days.
The measures of yesterday seem to have fruited so far. The graph shows the traffic over the last 2 weeks. Same period applies for the second graph showing the size of the media bucket. Will wait 2-3 days before changing back the retention time from 3 to 4 days.
This is the graph of the size of the media bucket over the last 3 months. Clearly visible when the scraping started. The drop in size at the end is me cleaning up after 3 instead of 4 days.
We are looking for the 403 status code to go up (403 means forbidden).
ℹ️ (increased) scraping ℹ️
Since roughly a week there is increased scraping taking place on mountains.social. This results in higher load on the server (which you might have noticed) and also an increase in media storage (as older posts are read more frequently, they are no candidate anymore for daily housekeeping).
I remove media for remote posts older than 4 days (they will be fetched again when someone needs that media again). The used media storage was stable around 90 GB, but has increased to 200 GB in the past week. Of course this will increase storage costs.
As a first measure I have changed the housekeeping to delete media from remote posts older than 3 days instead of 4. In the coming days I will have to setup blocking IP addresses / ranges from the seen scrapers, as the currently configured robots.txt ("please do no scrape") is just plainly ignored by them.
Will not write down what I think about them, but I guess you can figure (hint: it is quite explicit).
I have started to block the first IP-ranges (the most obvious ones).
Their traffic will still reach the webserver (consuming CPU), but will be dropped there. It should not reach the actual posts anymore (and therefore will also not touch their retention time. The cleanup of media from older remote posts will therefore be possible.
This is of course a cat and mouse game, so let's see how this evolves.
In case you notice negatives side-effects from your side, please give me a shout.
ℹ️ Wrapstodon ℹ️
For the people on mountains.social I have generated your Wrapstodon in case interested.
You will have to go to the web as you will find a notification there. Click on the link and you will find your stats.
Apparently meaning:
- the “lurker” is someone who posts, replies, and boosts less than a certain amount a year
- the “cool-hunter” is someone who boosts more than twice what they post
- the “pollster” is someone whose posts are at least 10% polls
- the “social butterfly” is someone who replies more than they post themselves
- the “oracle” is a person that does not fall into any of these categories
ℹ️ (increased) scraping ℹ️
Since roughly a week there is increased scraping taking place on mountains.social. This results in higher load on the server (which you might have noticed) and also an increase in media storage (as older posts are read more frequently, they are no candidate anymore for daily housekeeping).
I remove media for remote posts older than 4 days (they will be fetched again when someone needs that media again). The used media storage was stable around 90 GB, but has increased to 200 GB in the past week. Of course this will increase storage costs.
As a first measure I have changed the housekeeping to delete media from remote posts older than 3 days instead of 4. In the coming days I will have to setup blocking IP addresses / ranges from the seen scrapers, as the currently configured robots.txt ("please do no scrape") is just plainly ignored by them.
Will not write down what I think about them, but I guess you can figure (hint: it is quite explicit).
✅ maintenance ✅
A bit earlier executed than communicated, but now I was able to do this in the half time of the Handball Worldcup match France 🇫🇷 against the Netherlands 🇳🇱.
Release notes of version 4.5.3:
⚠️ maintenance ⚠️
I was not really planning to perform some maintenance, but then Mastodon version 4.5.3 came out, so I will combine that upgrade with the regular maintenance this evening on both mountains.social and mountains.photos.
ℹ️ mountains.community ℹ️
I run mountains.social and mountains.photos pretty much alone (supported by a loyal group of donors), which I like a lot. However, this also meant that I had to think about the situation where I would not be able to maintain the instances all of a sudden.
A couple of months ago I asked a fellow experienced Fediverse Admin to cover for me in that case and he was very willing to assist here (I will not disclose the name). He would gain access over all the passwords and documentation, so he could maintain it further, handover the instance or shut it down in case he desires (with the minimum of 3 months notice as written in the Mastodon Covenant that I "signed", so people could migrate to other instances).
Please rest assured that I have no intention to stop maintaining all of a sudden 😅 , but I wanted to inform you for the eventuality.
ℹ️ mountains.social ℹ️
Nice to see the built-in translation feature is getting used regularly. We are still able to stay in the free plan of the Deepl API. This gives us 500.000 characters per month to translate.
ℹ️ mountains.photos ℹ️
Have tried to optimize the PostgreSQL memory configuration. We'll see if this helps.
ℹ️ mountains.photos ℹ️
We still had memory issues in certain cases and since I had optimized everything that I could already, I increased the system just now. We are now running on a CX33 (4 CPU / 8 GB).
✅ mountains.photos ✅
The instance is back again. Seemed to be a memory issue of the PostgreSQL database. Will analyze later.
ℹ️ mountains.photos ℹ️
Have tried to optimize the PostgreSQL memory configuration. We'll see if this helps.
⚠️ maintenance ⚠️
Somewhere in the next hour I will upgrade mountains.social to version 4.5.2 (just a small bugfix release).
Will also perform some minor maintenance on mountains.photos.
⚠️ maintenance ⚠️
Somewhere in the next hour I will upgrade mountains.social to version 4.5.2 (just a small bugfix release).
Will also perform some minor maintenance on mountains.photos.
⚠️ mountains.photos ⚠️
There is an issue with the Pixelfed instance mountains.photos. Am working with Hetzner to get it fixed.
⚠️ mountains.photos ⚠️
There is an issue with the Pixelfed instance mountains.photos. Am working with Hetzner to get it fixed.
ℹ️ balance sheet October ℹ️
Did some accounting.
October balance sheet (and history) also at: https://mountains.community/behind-the-scenes#accounting
⚠️ maintenance ⚠️
Tonight at 21:00 CET I will be performing some maintenance on mountains.social. One of this things will be updating to Mastodon version 4.5.1. There most likely will some small downtime involved.