Hi Dan, At my project, we aren't that big on directory and user set up but the fact that services can be started and stopped automatically on server reboots and ensuring single running copies of the service is of significance. Now currently we are running Solr 4.4 but pretty soon we're going to upgrade to Solr 4.10.4. If I'm not wrong, the install scripts that set everything up ship with Solr 5.x. In this case, how do I set up my solr instances to behave like services created by the scripts in 5.x? I understand that this would entail setting up the init scripts and environment variables, and I do not have a lot of experience with those, so it'll be great if you can just walk me through it.
Or maybe it'll be simple enough to just manually follow the script from 5.x and adapt it to my solr instance as I go? Thanks. On Wed, Feb 17, 2016 at 11:40 PM Davis, Daniel (NIH/NLM) [C] < daniel.da...@nih.gov> wrote: > So, running solr as a service also runs it as a process. In typical > Linux environments, (based on initscripts), a service is a process > installed to meet additional considerations: > > - Putting logs in predictable places where system operators and > administrators expect to see logs - /var/logs > - Putting dynamic data that varies again in predictable places where > system administrators expect to see dynamic data. > - Putting code for the process in /opt/solr - the /opt filesystem is for > non-operating system components > - Putting configuration files for the process again in predictable places. > - Running the process as a non-root user, but also as a user that is not > any one user's account - e.g. a "service" account > - Making sure Solr starts at system startup and stops at system shutdown > - Making sure only a single copy of the service is running > > The options implemented in the install_solr_service.sh command are meant > to be generic to many Linux environments, e.g. appropriate for RHEL/CentOS, > Ubuntu, and Amazon Linux. My organization is large enough (and perhaps > peculiar enough) to have its own standards for where administrators expect > to see logs and where dynamic data should go. However, I still need to > make sure to run it as a service, and this is part of taking it to > production. > > The command /sbin/service is part of a package called "initscripts" which > is used on a number of different Linux environments. Many systems are now > using both that package and another, "systemd", that starts things somewhat > differently. > > Hope this helps, > > Dan Davis, Systems/Applications Architect (Contractor), > Office of Computer and Communications Systems, > National Library of Medicine, NIH > > > -----Original Message----- > From: Binoy Dalal [mailto:binoydala...@gmail.com] > Sent: Wednesday, February 17, 2016 2:17 AM > To: SOLR users group <solr-user@lucene.apache.org> > Subject: Running solr as a service vs. Running it as a process > > Hello everyone, > I've read about running solr as a service but I don't understand what it > really means. > > I went through the "Taking solr to production" documentation on the wiki > which suggests that solr be installed using the script provided and run as > a service. > From what I could glean, the script creates a directory structure and sets > various environment variables and then starts solr using the service > command. > How is this different from setting up solr manually and starting solr > using `./solr start`? > > Currently in my project, we start solr as a process using the `./` Is this > something that should be avoided and if so why? > > Additionally, and I know that this is not the right place to ask, yet if > someone could explain what the service command actually does, that would be > great. I've read a few articles and they say that it runs the init script > in as predictable an environment as possible, but what does that mean? > > Thanks > -- > Regards, > Binoy Dalal > -- Regards, Binoy Dalal
