What is a decent approach to search indexing servers from workstations?

At work I will be installing a full file server to replace an aging off-the-shelf NAS appliance. Users will connect through NFS and Samba from both Windows and Linux desktops. There are thousands of files on the old NAS and through 10 years, folder organization has grown out of control. I will be reorganizing the directory structure, but finding a file among thousands is a challenge. Hence, a desire to index the files.

From what I read, I can configure each individual desktop to index the file server mapping. Seems clunky and inefficient.

I looked at some indexing apps. I am not seeing how to centralize the indexing on the server such that the workstations use that index rather than a local index.

How are you gurus doing this?

Whilst we wait for a guru ...

I would simply use ssh to run locate on the server. Quick and efficient. But there may be all sorts of internal issues in a business. Especially with the Win users .... :shrug:
Else maybe after running the updatedb (on the server) push the entire list out to a webpage - let the user just search as normal from their browser.

All the above presumes everybody (internally) is authorised to see all the filenames.

Yeah -- running updatedb on the server does not help client workstations. Especially Windows workstations.

There are indexing systems such as solr and elasticsearch, but they seem designed for Big Stuff. And I have not found anything helpful how client workstations access those indexes other than a web browser.

I looked at Recoll, which can run on Windows as well as Linux, but I don't see the connection to get client workstations to use the server's index.

What's wrong with ssh - and maybe something like plink or it's ilk for Win ?.

Set up a common script name, no-one will know the difference.

This needs to be pointy-clicky. The non technical office users are not going to use SSH on a Windows system. They just want to find files.

Friendly bump.

Recoll is one that can run locally. It will run on several operating systems but I'm not sure if you can use a shared index. It's most likely that each index must be run individually.

For web interfaced indexes, there used to be Swish-E and ht://dig but they are abandoned. There are several Lucene-based ones like Solr and Elasticsearch. However, those are both Java as is the Lucene engine underneath them.

Recoll has a web UI (https://github.com/koniu/recoll-webui/) which can give access to a central index. It's not designed for managing access rights though, just for a common datastore. This may be a problem or not for your use case.

jf

I will investigate. Thanks.

Thanks. Browsing the docs shows that multiple indexes are supported. As both Linux and Windows see a network share as a local directory, I wonder if I can generate indexes on the server and then configure Recoll on the workstations to use that index. Looks plausible but I'll need to test.