The tarball installation provides a self-contained directory with everything you need to run OpenSearch, including an integrated Java Development Kit (JDK). The tarball is a good option for testing and development.
I’m not planning on installing via Docker and just plan on creating a cluster from multiple VM’s. My question is… Since the docs say that the tarball option is good for testing and development, how would I go about this if I wasn’t planning on this being “testing and development”?
This cluster is not going to some super important “mission critical” cluster. But I’d consider it more “prod” that “testing and development”. It will have multiple master nodes, multiple hot/warm nodes, and multiple coordinating nodes. So it gave me pause to see that the tarball install is only good for “testing and development”.
Is there a separate install method for a “prod” deployment that I haven’t found yet? Or is it just because the tarball comes with it’s own JDK?
Some clarification on the statement “the tarball is a good option for testing and development” would be helpful.
The wording here is pretty specific - it’s good for testing and development, but I don’t think it says it’s bad for production. In your situation, go for that the tarball.
A easier option for a lot of folks in production is package manager installs (DEB/RPM) but those aren’t yet available (next version). I think once those are out, that will be a better recommendation.
out of interest: what will these contain? with the tarballs we have various distributions available (“full” vs. “min”; hopefully soon also with “no-jdk”). will the DEB/RPM ship as a “full” package with its own JDK or will there also be multiple DEB/RPM packages available (one for each distribution)?
i couldn’t find any information on this in the tickets.
ok, so in a docker image built for a specific purpose we’d still use the tarballs instead of the RPM packages? (i’m fine with that, it’s what i’m currently doing)
@saeed.kazemi - I think the docs are pretty good for configuration - although not really a tutorial - still a great resource.
I’m installing from the tarball also but I’m automating it with Ansible. And while there is an Ansible install method but it didn’t really suit my needs. So I’m using the existing Ansible install method as a starting point and making modifications from there. I mention this to you because the Ansible playbook they made is a good way to see how to configure a basic setup. It might be helpful to you.
Hi, one thing that was not clear to me is if after running the install script ./opensearch-tar-install.sh or ./bin/opensearch was suppose to free the terminal, because in my case it seems that it doesn’t create a daemon that runs on the background. If I exit the terminal, it stops the service.
I tried also using jdk outside of the tarball and passing an environmental variable to JAVA_HOME, but the result is the same
That is the correct behavior – OpenSearch itself is not a daemon and runs in the foreground. If you are not using systemd to manage the service, you need to run it using ‘nohup’ to keep it from dying when you close your terminal:
> nohup bin/opensearch &
all output will be captured in the ‘nohup.out’ file, unless you specifically redirect stdout somewhere else.