Yuka

This is a kernel bug, but glibc is strictly a userspace library. If you need to work around a glibc limitation, you can compile the code with `-static` on a glibc platform.
It bundles glibc directly into the executable, letting you run it on non-glibc systems without worrying about shared library dependencies.
Maybe ...
gcc exp.c -o exp -static -lutil
incus file push exp toys/tmp/exp
?

2026-05-07 22:02:04 CEST

2 months late. But at least now I have a job.
#note1v32…t0ll

2026-05-07 21:53:44 CEST

Even though I used to really like spicy food, now my stomach can't even take a little bit of it.

2026-04-08 12:10:10 CEST

Contabo is ridiculous. They said s.umeyashiki.org is used to distribute video files containing infringing content exclusively owned by the Rights Owners.
The complaint comes from Vobile Compliance. They said they act on behalf of Paramount Global companies CBS Broadcasting Inc., CBS Studios Inc., Paramount Pictures Corporation, Showtime Networks Inc., Viacom International, Inc., Black Entertainment Television LLC, and other Paramount Global subsidiaries and affiliates.
https://static-s.umeyashiki.org/media/ffd3c5b7b975602b702b3ce5afaa7cb91b97b5cbcf489000f71eeedfe4188060.png

2026-03-22 21:05:50 CET

- reply

I also failed to achieve my goals, but if I don't move forward, the burden will get bigger everyday.
I am still positive with my life and will always be (hopefully). I believe most of those people on the top have experienced worse life situations than me.
I create new goals and continue my battle.
If stop here, I will never achieve goals forever. That being said, it is a failure guarantee.
If I do my best, there is a chance for a better life, despite it is not guaranteed.

2026-03-22 20:55:05 CET

- reply

Fortunately, my family still takes care of me. But I can't rely on them forever. I hope I can be independent this year.
What do you do for living?

2026-03-22 20:34:31 CET

I should learn more skills to get a job easily.

2026-03-11 13:07:06 CET

Today I realized that my Linux Contabo VPS has a built-in script in `/etc/cron.hourly` that drops the cache every hour. I found it when I checked dmesg; there are so many "drop_caches: 1" lines.
I guess the memory is not dedicated, but shared via KVM memory ballooning.
It seems Contabo tries so hard to reclaim physical memory from its customers by dropping page caches every hour. I feel like Contabo is somewhat notorious for aggressive overprovisioning.
I will delete the `free` script.
#linux #contabo #mm #kvm #vps
https://static-s.umeyashiki.org/media/f3a5c8cc3960b237590df91ccfbbfa6ad897446563c1da9a413a14f0fb7ef6ec.png
https://static-s.umeyashiki.org/media/7c1d69ddf7271e2903b4b8ef30b14422c4fcd7bc0a0615781a95b6810ce7f6f1.png

2026-03-10 19:53:37 CET

- reply

Discord :awa:

2026-03-05 23:23:51 CET

- reply

make it run inside docker part :blood_sob:

2026-03-05 12:32:12 CET

I still fail to set these things up.
#note1csx…tg6y

2026-03-04 01:58:32 CET

- reply

you still didn't answer the question. The user asked "what country" 😅

2026-03-03 04:40:46 CET

- reply

Can you share your DB backup command?

2026-03-03 04:40:03 CET

- reply

Sorry for a stupid question. But what is index data?

2026-03-03 04:34:35 CET

- reply

my pg_dump already preserves the indexes. Did I miss something?
https://static-s.umeyashiki.org/media/43d26d3025a8da02b8c4550856758e708bd90a4a14611b64f8be1db09f569804.png

2026-03-03 03:35:37 CET

I expect to get a job this March.

2026-03-03 03:13:06 CET

My new daily backup script, pg_dump with zstd compression level 19.
```
docker exec \
ak-postgres-1 \
pg_dump -U umeyashiki umeyashiki_akkoma | \
nice -n 19 \
ionice -c 3 \
chrt --idle 0 \
zstd -T0 -19 --rsyncable -q > "$BACKUP_DIR/db_latest.sql.zst";
```
Today I learned something new about the scheduling priority. Since zstd compression is very CPU-intensive, set it to low priority so it doesn't slow the entire system down during compression.
Commands that precede zstd here are:
- `nice -n 19 [cmd]`
- `ionice -c 3 [cmd]`
- `chrt --idle 0 [cmd]`
By chaining nice, ionice, and chrt together before the zstd command, the script forces the compression process to run with the absolute lowest possible priority for both the CPU and the disk.
- nice 19 is the lowest priority CPU priority.
- ionice with class 3 means idle. A program running with idle I/O priority will only get disk time when no other program has asked for disk I/O for a defined grace period.
- `chrt --idle 0`: Set scheduling policy to `SCHED_IDLE` (scheduling very low priority jobs).
References:
- `man 1 chrt`
- `man 1 nice`
- `man 1 ionice`
#linux #docker #zstd #postgresql #sched

2026-03-03 02:52:16 CET

My Akkoma and PostgreSQL are now running inside Docker. I guess the mail server can also run inside Docker.
```
# docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
0e70c1dc246f akkoma:latest "/opt/akkoma/docker-…" 3 seconds ago Up 2 seconds 4000/tcp ak-akkoma-1
2ba1820747a9 pl-postgres:18-rum "docker-entrypoint.s…" 3 seconds ago Up 3 seconds 5432/tcp ak-postgres-1
```
#linux #docker #akkoma #postgresql

2026-03-02 19:15:11 CET

- reply

perfect reminder, thanks.

2026-03-02 19:12:11 CET

- reply

I will use btrfs it it will solve🫠

2026-03-02 19:11:17 CET

- reply

Will that solve the problem you mentioned?

2026-03-02 19:09:12 CET

- reply

Can I mount another fs into docker? Or is it strictly required to use overlayfs as the storage?

2026-03-02 18:34:05 CET

- reply

I want to know the reason. Why is that?

2026-03-02 18:18:15 CET

Someone suggested me to learn docker and move my running services into it.

2026-03-02 07:54:34 CET

- reply

what happened?

2026-03-02 07:12:54 CET

- reply

What I expect after "apt remove" -> "apt install" is that I get new default config files.
But that didn't happen; instead, the old, broken config files remained.

2026-03-02 07:11:26 CET

- reply

I have attempted to set up Postfix and Dovecot on my old VPS, but I think I made a mistake in the configuration.
When I want to repeat the process from the start, I tried to "apt remove" Postfix and Dovecot, but several old config files were not properly purged by "apt remove".
I ended up using broken configurations. I was burned out, then decided to skip it. Now I have a freshly installed server. I hope I do not make the same mistake.

2026-03-02 06:58:33 CET

Another attempt to install the mail server. I have failed twice. Must be successful today.

2026-02-28 14:41:41 CET

I have zero preparations to deal with WW3.

2026-02-28 14:08:36 CET

- reply

Nooo...
Why would someone want their own server to crash? 😭
I want stability, but that is not my priority as my financial condition doesn't support it. I wish Contabo were stable and my projects didn't crash because of Contabo.

2026-02-28 13:58:51 CET

- reply

Sure, I know what I have. This is just a hobby project; having a proper life is my priority 😭

2026-02-28 13:52:51 CET

- reply

> good luck
Thanks!
> what kind of service you will run here?
What I have planned: Blog, Mail server, web scraper, and Discord bot.
There are more in my backlog. But before that, I'll think about getting a bigger VPS. That means I should first think about how to earn more income.

2026-02-28 12:21:52 CET

- reply

I will not run out of things to do in the near future, but will run out of resources soon 😭
Akkoma is just one of N services I am going to run on this VPS.
There will be more services very soon. It's not like I have multiple servers to tinker. This VPS is not purely used only for Akkoma, I will do everything else in this one server.
As a preparation, I'll at least spare a 6 GB partition for the swap. Resizing my btrfs should accommodate it.

2026-02-28 12:06:58 CET

- reply

If the swap isn't touched, there should be no extra disk I/O for swapping. How does that affect disk performance?

2026-02-28 12:02:52 CET

- reply

Why does 12 GB of RAM not need swap?
I'm thinking of a sudden spike in memory pressure; it could actually fill up.

2026-02-28 11:28:44 CET

- reply

On server.

2026-02-28 11:23:02 CET

I forgor to spare a partition for linux swap 😭

2026-02-28 11:06:08 CET

- reply

I need to consider a few things before integrating your mediaproxyoma, so this is on the back burner for now.

2026-02-28 10:56:50 CET

SPF, DKIM, DMARC
What kind of creature are those?

2026-02-28 10:53:01 CET

- reply

Not quick. I don't understand the table structures in your tutorial. I have several questions, but it'll be a bit long to write them now. Not mood.

2026-02-28 10:48:02 CET

I don't feel any slowdown after migrating to the new server, despite the link speed being only 300 Mbps.
It simply because 300 Mbps isn't actually fully utilized to just run this tiny Akkoma instance.

2026-02-28 10:45:47 CET

- reply

ok, thanks. I've created a subvolume specific to pgsql data and set +C (nodatacow) to the directory.
```
root@vmi3114076:/home/ubuntu/ak# lsattr akkoma/pgdata/
---------------C------ akkoma/pgdata/18
root@vmi3114076:/home/ubuntu/ak# btrfs subvolume list .
ID 256 gen 142 top level 5 path home/ubuntu/ak/akkoma/pgdata
```
Next TODO is the mail server, but it's not my priority. Maybe I'll do it next week.

2026-02-28 10:33:14 CET

- reply

> DON’T FORGET TO DISABLE CoW ON PSQL DB
Why?

2026-02-28 10:30:18 CET

- reply

migrating from ext4 to btrfs feels easy when I have a fresh installed server, because I don't worry about losing data if I make a mistake with the fdisk command 😆
https://static-s.umeyashiki.org/media/df22ff6c90633f9025d11751c3fea48ef3fa4158d9d02e27bcebca14bc16a132.png

2026-02-28 10:26:38 CET

I'm back!

2026-02-28 08:44:47 CET

- reply

https://www.speedtest.net/result/c/a6feff77-e9f7-4e04-9470-ba0ed5243808
I should survive with this. I will upgrade gradually as my income increases (not sure when).

2026-02-28 08:36:13 CET

- reply

That's much slower than Vultr 😭

2026-02-28 08:35:40 CET

- reply

I forgor, I think it's 300 Mbps.

2026-02-28 08:23:52 CET

- reply

I've calculated everything, it's cheaper to commit to 12 months.

2026-02-28 08:20:35 CET

Checked out a slightly bigger VPS on Contabo for $6.36/month with a 12-month contract period.
This should be the last week I use Vulr. I will move everything to Contabo.
Thanks for the recommendation.
https://static-s.umeyashiki.org/media/99819356cf99ab53a68bfadca0617d6b21da95f79b04403a39c8c325148132d9.png

2026-02-28 06:02:39 CET

- reply

I'm poor, don't have many choices 😭

2026-02-28 06:01:04 CET

- reply

I'll check Contabo, thanks.

2026-02-28 05:47:21 CET

- reply

gamble? What do you do exactly?

2026-02-28 05:17:02 CET

- reply

I can't bear the electricity and server room costs. I prefer renting VMs on the cloud.

2026-02-28 05:12:20 CET

I want to buy a bigger VPS, but no money.

2026-02-27 06:34:46 CET

People here use the `#introduction` tag to introduce theirselves. Maybe I should do too.

2026-02-26 07:22:08 CET

- reply

I copied the config from Pleroma upstream.
https://git.pleroma.social/pleroma/pleroma/src/branch/develop/installation/pleroma-mediaproxy.nginx#L85-L96
It still needs to communicate with the Elixir backend, but the content caching is handled by nginx.

2026-02-26 07:18:51 CET

- reply

Oh, it turns out nginx already has `max_size` parameter, I will adjust that. Now, I am not worried about cache size blowing up, as it's already constrained in size.
https://static-s.umeyashiki.org/media/82512f61d403bd37c9e7e324194abde98744d0383a2ccf35fcab9955a39a9652.png

2026-02-26 07:11:43 CET

- reply

What varnish will do in addition to the nginx proxy cache?

2026-02-26 07:09:58 CET

- reply

I'm not sure if it has a mechanism to limit the cache size. Don't have a strong motivation to look for the answer.

2026-02-26 07:08:29 CET

- reply

maybe not, I will purge them as they grow big.

2026-02-26 07:06:57 CET

- reply

it's around 1G (nginx logs and proxy cache)
```
# du -hd0 pleroma*
185M pleroma-logs
929M pleroma-media-cache
```

2026-02-26 07:04:21 CET

- reply

seeing your size, it looks like I still have big enough room to grow.

2026-02-26 07:02:37 CET

- reply

> are you using cache?
Mediaproxy cache? Yes.

2026-02-26 06:39:08 CET

Umeyashiki Akkoma takes 3.5G of disk space, with 2.7G being the PostgreSQL data. I hope the size doesn't go up quickly enough to its full capacity before I can afford to buy a bigger VPS.
https://static-s.umeyashiki.org/media/874c9dcdfdb75f451703e97647d113e36e39e1140e1ef8090bb80714aabcb08e.png

2026-02-25 16:55:50 CET

[40165.176024] TCP: request_sock_TCP: Possible SYN flooding on port 443. Sending cookies. Check SNMP counters.

2026-02-25 05:37:22 CET

- reply

mornyaaan, send cat pics please

2026-02-25 04:59:40 CET

- reply

what's that?

2026-02-25 04:53:13 CET

Something feels a bit scary about his knowledge. I was a bit embarrassed to be called wrong 😭
#note18vv…8mhj

2026-02-25 04:16:26 CET

- reply

with clang, it explicitly mentions it is a GNU extension.
https://godbolt.org/z/ff95csq66
```
<source>:7:26: error: use of GNU ?: conditional expression extension, omitting middle operand [-Werror,-Wgnu-conditional-omitted-operand]
7 | p = realloc(p, len) ?: p;
| ^
1 error generated.
Compiler returned: 1
```

2026-02-25 04:06:48 CET

- reply

b-b-bbut that's a GNU extension, it's cheating 😭
That won't work with -pedantic-errors. https://godbolt.org/z/1qP5aMeGT
<source>: In function 'fx':
<source>:7:26: error: ISO C forbids omitting the middle term of a '?:' expression [-Wpedantic]
7 | p = realloc(p, len) ?: p;
| ^
Compiler returned: 1

2026-02-25 01:47:59 CET

You can't safely realloc() without a temporary variable.

2026-02-24 11:45:57 CET

- reply

You're so cool. I wish my brain could understand that kind of complexity. Even today, I still can't configure my VPS correctly. My ability is still far away from configuring Akkoma to that level 😭

2026-02-24 11:12:12 CET

- reply

Wooaah... smart optimization. I wouldn't have thought about something like that.
It took me some time to understand the pull request as there is no provided description in the PR. I can understand it now. That makes sense for local media uploads because the file name must be unique for that file, and it never changes forever.
FWIW, I've applied the immutable cache-control hdr for static-s.umeyashiki.org. I didn't change the Akkoma sources because it's purely served by nginx (except for the media proxy, I left it as is for the time being).
After adding the following directive in the nginx conf:
`add_header Cache-Control "public, max-age=1209600, immutable" always;`
And then after purging the Cloudflare cache, I verified it with curl that the header now contains "immutable". I hope it speeds up the load on my browser. Though I don't actually understand how it works behind the scenes.
Arigatou gozaimas
https://static-s.umeyashiki.org/media/460d641258ccf33389829b173686812ce6105d1fe7f7d0e5ea00fd94316c11e1.png

2026-02-24 10:05:14 CET

- reply

nginx 1.29.6 also supports QUIC, but I don't enable it because it runs behind Cloudflare. Cloudflare uses H2 to communicate with the origin server, sadly no QUIC in transit, only the CF's frontend part uses QUIC.

2026-02-24 09:49:18 CET

- reply

Eeeeh, I didn't expect you'd actually write it for me.
I will try to follow your tutorial blog. Arigatou gozaimas

2026-02-24 09:45:17 CET

I'm accessing s.umeyashiki.org via QUIC. Thanks to Cloudflare.
Protocol h3 = QUIC = HTTP/3
https://static-s.umeyashiki.org/media/d20402f3e393fe1e5e68356b7aecea09cb984b2d54f5fc19899575b2fa14a51e.png

2026-02-24 08:27:03 CET

Alien's shell script 😭
#note14pf…893w

2026-02-24 08:09:14 CET

- reply

yaabeee, your post confused me. I thought I posted that pic. Turns out that's a custom cover for CW. My brain lagged a bit

2026-02-24 07:55:55 CET

- reply

I could understand the Digital Ocean tutorial blog format I sent. The steps are very clear. But it might've been outdated, which resulted in a different result on my VPS.
I just realized that the blog I referenced was written in 2013. Maybe it's no longer applicable in 2026. 😭

2026-02-24 07:42:31 CET

- reply

I checked your repository, but I don't understand the explanation in the `README.md` step 3. I'm not so good at inferring the last part. I honestly don't know what to do with those config files 😅

2026-02-24 07:32:11 CET

- reply

I followed this blog https://www.digitalocean.com/community/tutorials/how-to-set-up-a-postfix-e-mail-server-with-dovecot
It was a half-complete setup as I was exhausted from configuring the Postfix part. I didn't remember exactly where I stopped, but the result on my VPS was not the same as mentioned in that blog.
It's still a mess on my VPS rn, so I decided to stop the Postfix service instead of leaving it running. Wasting too much time :blobmeltsob:

2026-02-24 07:09:49 CET

Just lost several hours of my life to silly CORS issues while setting up https://static-s.umeyashiki.org as my Akkoma media base URL :blobthump:
Nginx "gotcha": header inheritance.
If you put an `add_header` directive inside a location block, it completely suppresses all `add_header` directives defined in the parent server block.
```
server {
add_header "X-Hdr-A" "Value";
add_header "X-Hdr-B" "Value";
location / {
add_header "X-Hdr-C" "Value"; # <-- This wipes out A and B!
access_log /var/log/a.log;
}
location /bb {
access_log /var/log/b.log; # <-- A and B will show up here just fine.
}
}
```
If your global headers are mysteriously disappearing on certain paths, maybe check your location blocks.
#http #nginx #cors #akkoma

2026-02-17 00:55:27 CET

- reply

But I haven't enabled the email functionality yet. Attempted to configure a mail server, but it was hard, and I ended up throwing it away.

2026-02-17 00:53:19 CET

Finally done setting up this Akkoma server.

2026-02-17 00:51:06 CET

Hello!