I’m no good with Docker, but what I can do is pool funds and rally people to get it done.

It’s totally fine that you only support some apps — but when it includes core functionality for an app this is the sort of thing people are going to want.

Ok, I’ll start a thread about crowdfunding ElasticSearch packaging for a custom Cloudron docker image.

They already said it's not on the immediate roadmap.

Yeah, but they have also previously sais somewhere on here the plan is to make an addon, plus @girish liked my post above about different versions perhaps not really being an issue

If Cloudron isn’t going to work on it,

They already said it's not on the immediate roadmap.

so I can figure out how to package it manually.

Everyone would appreciate your efforts in figuring out how to package it manually and sharing that! Thank you!

If Cloudron isn’t going to work on it, would be good to know, so I can figure out how to package it manually.

@jdaviescoates said in How to install Elasticsearch on Mastodon?:

I note that the Mastodon docs do now say:

Mastodon is tested with ElasticSearch version 7. It should support OpenSearch, as well as ElectisSearch versions 6 and 8, but those setups are not officially supported.

So sounds to me like the

@girish said in How to install Elasticsearch on Mastodon?:

apps couldn't agree on which ES version to use

Might not be much of an issue in practice.

It'd be great to have an ElasticSearch Addon added to Cloudron.

What is the recommended path here that doesn’t mean running a second machine?

@girish

Seems the Debian host I was using on Linode (using "Secure Your Server" image) was blocking port 9200 by default so had to adjust iptables to allow external access. Had a hard time tracking that down.

Added:

iptables -I INPUT -p tcp --dport 9200 --syn -j ACCEPT

then saved it. Worked immediately.

I installed Cloudron and attached a VLAN. Then, from a web terminal of an app

root@3270ef46-f6cc-463a-bff2-33062223f536:/app/code# ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host 
       valid_lft forever preferred_lft forever
23: eth0@if24: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default 
    link/ether 02:42:ac:12:13:23 brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet 172.18.19.35/16 brd 172.18.255.255 scope global eth0
       valid_lft forever preferred_lft forever
    inet6 fd00:c107:d509::8/64 scope global nodad 
       valid_lft forever preferred_lft forever
    inet6 fe80::42:acff:fe12:1323/64 scope link 
       valid_lft forever preferred_lft forever

I ran nc -l 3000 on another server which is also in same LAN. Then, I can connect from app terminal:

root@3270ef46-f6cc-463a-bff2-33062223f536:/app/code# telnet 192.0.2.3 3000
Trying 192.0.2.3...
Connected to 192.0.2.3.
Escape character is '^]'.

If I understand correctly now: you put your ES and Cloudron on the same VLAN. You are able to ping the ES from Cloudron host machine but unable to do so from the web terminal. Correct?

When I use the Cloudron Terminal under my Mastodon app, it does not show the eth1 interface, only loopback and eth0.

So it is not my host.

Outcome is, Cloudron can't see the VLAN network on Linode - at least nothing is showing in the UI, and I am unable to use ping in the Cloudron terminal for some reason.

I'll keep testing.

For me on Linode, all I did extra was to create a VLAN and connect eth1 on my Cloudron to it, as well as eth1 on my Elasticsearch server. Then used this private link to connect the two.

Edit : Actually damn while troublshooting why I was having an issue getting it to work, it seems Cloudron does not recognise multiple NIC/eth interfaces. So I can't have eth1 on a VLAN with a 10.0.0.x/24 address.

See Linode setup here: https://www.linode.com/docs/products/networking/vlans/get-started/

Cloudron via "ip link show" isn't showing the eth1 interface at all. @girish any way to fix this toi make it work?

Edit 2: Make sure to troubleshoot the network.host on the Elasticsearch server and port 9200 is open in iptables - mine was refusing connections for some reason. Working on VLAN now.

@nichu42 I was just looking at it briefly. In case you go down this route... For some reason, they only support ES7 per https://github.com/mastodon/mastodon/issues/26544#issuecomment-1684005007 which is almost https://www.elastic.co/support/eol .

Also, Nextcloud ES integration now requires ES8 which could conflict with the previous ES7 requirement.

Finally, https://github.com/mastodon/mastodon/issues/18535 suggests OpenSearch 2.4.x might be supported.

I note that the Mastodon docs do now say:

Mastodon is tested with ElasticSearch version 7. It should support OpenSearch, as well as ElectisSearch versions 6 and 8, but those setups are not officially supported.

(also, I've just submitted a pull request to fix their tiny Elastis typo )

For the time frame, when will Elasticsearch become avaible or not

As @girish said it's not yet on the roadmap, so I don't think it'll really be possible to guesitmate the time frame.

@redegelde said in How to install Elasticsearch on Mastodon?:

i understand this is needed with the 4.2 version

To be clear, it is not required by 4.2 (as in you can use 4.2 without it). But yes, it can be used with 4.2 and the Cloudron package for 4.2 is already available.

Also, Nextcloud ES integration now requires ES8 which could conflict with the previous ES7 requirement.

Finally, https://github.com/mastodon/mastodon/issues/18535 suggests OpenSearch 2.4.x might be supported.

Let me if you come up with something!