Assaf, If you want to use Drill as a Server vs Embedded mode, you may be better off to install ZK (Zookeeper) on the node (assuming you run Linux) and then just run Drill as a single node cluster/distributed mode. Opens up your options to scale to multiple nodes easily in the future as well. Apache ZK in pretty simple to install and relatively lightweight.
Using screen, etc to simulate a server with embedded mode is just asking for problems later when providing a service to multiple users. You may been want to consider a full cluster deployment with DFS underneath, even if it is just a single node to start. —Andries > On Oct 27, 2015, at 8:49 AM, Kristine Hahn <[email protected]> wrote: > > I don't know what you mean by "using screen". You cannot use the wildcard > asterisk in the query afaik. You can query a directory of files, but I > believe the files need to meet the conditions mentioned in > http://drill.apache.org/docs/querying-directories/. I could not query a GZ > file unless I used a storage plugin configuration based on the dfs default > one that specified the GZ format for JSON as described in > http://drill.apache.org/docs/querying-plain-text-files/#querying-compressed-files. > Creating a storage plugin configuration (or modifying an existing default > configuration, such as dfs) is tricky. The configuration is wiped out when > you click UPDATE if it isn't valid JSON, so save your attempt somewhere so > you don't have to retype stuff . Valid configurations are not saved by > default in embedded mode. You have to modify the drill-override.conf (see > http://drill.apache.org/docs/storage-plugin-registration/#storage-plugin-configuration-persistance). > Also,please take a look at > http://drill.apache.org/docs/plugin-configuration-basics/. Sorry Drill info > is not organized better. > > Kristine Hahn > Sr. Technical Writer > 415-497-8107 @krishahn skype:krishahn > > > On Tue, Oct 27, 2015 at 1:55 AM, Assaf Lowenstein <[email protected]> wrote: > >> Hi, >> >> I'm trying to setup a drill server, running in embedded mode, so users will >> be able to access a centralized location and I need assistance/confirmation >> on the following: >> >> 1. As it's a server, I need it to run Drill as daemon or other similar >> method. Everything I tried didn't work except using *screen. *Please >> confirm that using screen is ok and functionality doesn't regress as a >> result. >> 2. When trying to run >> >> create table dfs.tmp.`tableName` as select * from >> dfs.`/mnt/drill-work/XXX/XXX/*.gz`; (gz holds json files) >> >> I'm getting: >> >> org.apache.drill.common.exceptions.UserRemoteException: VALIDATION ERROR: >> From line 1, column 48 to line 1, column 91: Table >> '/mnt/drill-work/XXX/XXX/*.gz' not found [Error Id: >> 87c77937-c3c9-4d42-80c3-20d7de6738eb on XXX.XXX.com:XXX] >> >> >> the files definitely exist in /mnt/drill-work/XXX/XXX/*.gz. >> >> Question is, does drill go to the server /mnt/drill-work or does it look >> for it in on the client (laptop)? I'm using the web UI interface. >> >> Thinking that maybe Drill is looking at my laptop and not on the server, I >> was trying to change dfs storage definitions but I noticed that the update >> button doesn't work (!) so effectively I cannot make changes to the >> storage definitions. >> >> Thanks a lot! >> >> >> -- >> *Assaf Lowenstein*, *QA Manager, Totango* >> http://www.totango.com | Contact me @ cell: +(972) 052-4235550 >>
