nitrolife

and continuing my thought, overwatch was literally what you wanted in them first years. But it turned out that the main game group was not interested in playing for healers or tanks without damage, and as a result, the bias in choosing these roles was more than 3 times. This increased the waiting time in the queue quite a lot. As a result, everything returned to the option where each role can cause damage.

Mercy so easy only in low ranks.

If you have good move you can do teamkill by mercy under ultimate skill… Especially if your DPS no have damage. In overwatch you should kill by any role, but in healer mode you should don’t die at first, heal you team at second and do damage in third. That’s true less for mercy, because she has damage increase bean and you should change to it for support ultimates for example, but more on Birgit or for new healers.

And with tank too. You should protect your team, know all mechanics and do damage for enemies. For example sigma, you can eat Moira orb and Zarya ultimate skill by bullet protection skill. that is literally black hole. And if tank don’t do that you team in pain. And if you don’t damage targets you team lose.

But Overwatch bad example because that more arcade shooter. Many moves depend by skills, not for headshots. You literally have melee characters without any targeting.

I used lvm with SSD cache few years, but time to time I have problems with loads after reboot. If forgot about reboots all work great with LVM raid + LVM cache. Cache can be configured without raid. And you can add or remove cache in any time. Docs: https://man.archlinux.org/man/lvmcache.7

Topic starter don’t use docker containers. He use lxc containers. LXC is virtual machines on host kernel. Case is that: He have 3 virtual machines with shared disk. In 3 VMs have 3 different applications. Applications work under different users and can’t rewrite data on shared disk. That is user permissions problem.

the idea is that: all your applications work under the same user. or at least under the same group. because this is exactly how the differentiation of rights is applied.

A good plan is to create some kind of user in all three containers and run qbittorrent, samba and the third application under it.

A bad plan is to run everything under a random user with 777 rights, but this is a really bad plan.

Create a user in all three containers and work under it. That is not hard. Run qbittorent with that user. Config will be there: /home/user/.config . Then set that user for samba. I don’t know third app , but I think you can find how change user in manual.

You can use Revolt. Literaly Discord clone. https://github.com/revoltchat

You need to create an MX record in the DNS zone of your domain. Something like:

@ IN MX 10 my.zome.
@ IN MX 20 server1.my.zome.

You can create 1 MX record or more. 10 and 20 server priority for input mails.

Then you need to create an spf record. There are several options here. For example

@ IN TXT "v=spf1 +a +mx -all"

allows you to send emails from A domain records, then from MX domain records and prohibits from all other hosts.

Theoretically, you can only create an SPF record with A but without MX and dont create MX DNS records. Although I have not tried this configuration.

This is the minimum set after which you will get into spam, but at least the letters will reach.

You also need to make a PTR record to avoid spam folder, but this is not possible on a dynamic IP.

interesting facts about LVM:

1. You can make a volume snapshot of the system before a major change (for example, an update).

2. You can enable caching and use HDD together with SSD cache

3. You can build raid 0,1,5 directly on LVM (you still need modules from mdraid)

4. Even without a raid, you can expand the partition beyond one disk to another or migrate the partition from disk to disk (without even disabling it)

However, all this is done from the console and I do not know if there is a GUI.

I think that in order to solve such a question, we first need to consider something else. Why, if votes are so important to you, can’t you just create a bunch of accounts and vote honestly on any server?

As soon as we are really sure that 1 person is 1 vote, and not 10, 100, 10000 or any other number, then it is already possible to build trust checks between servers. Although it seems that this has not been solved even by large social networks.

The answer to your question in general is this: store the votes by servers and then double-check the result randomly.

S returns: 50 votes for a post from server A, 30 for a post from server B, 10 for a post from server C, etc. Then you can randomly check on these servers whether the amount is correct. However, there is no way to check the voices of server S, so they either have to be thrown out or still trust the server at its word. It is possible to fully verify server S only if registration on all servers goes through a trusted intermediary.

If you need fast online share you can use nginx + WebDAV module . You can use ip filters or auth. Or disable any auth.

Show all instances subscribers technically impossible. Or all instanses must go to all another instances (with banned too) and calculate total number. See local subscribers better then nothing.

That number show local subscribers. I think that not a bug.

You can create graph on the UI and then export to json. To be honest, it’s hard for me to imagine a situation where graphs need to be edited so often. After all, there are excellent template engines in Grafana itself. For quick look anyone can use Explore panel. Maybe I don’t have so much data…

Grafana sends an email screenshot of the graph when an event is triggered on the graph. You can see alerts part on any graph for understand.

You can configure grafana without gui. That explain in https://grafana.com/docs/grafana/latest/administration/provisioning/

prometheus use own time series database. you can connect influxdb to grafana and send alarms from grafana, but alertmanager better i think. node-explorer can collect all this data (sensors, VM/PC load etc.)

I use:

1. Monitoring server - prometheus
2. Alert manager for prometheus - alertmanager. You can write any triggers here.
3. Web UI for prometheus - Grafana
4. Exporters for prometheus - node-exporter, blackbox-exporter, mysql-exporter, psql-exporter etc. You can find exporter for everything you need.
5. Some services native support pormetheus. Docker for example: https://docs.docker.com/config/daemon/prometheus/

If you whant cluster you can install thanos on prometheus.

Try running top and see the global cpu/mem statistics when loading images. Look at what a high-load process is. Check that docker is not installed via snap.

Do I need to allow docker to use more than 6 GB?

Check exactly how much RAM NextCloud daemons are used. How much CPU used?

In general, it looks like an overload that occurs when NextCloud starts processing newly loaded files.

Unfortunately, everyone’s choice of one server will lead to centralization again. Therefore, it is much better to sort at least the top 5 servers and select them randomly in applications by default.

UPD: In addition, it will distribute the load on several nodes, which will have a good effect on performance.