These are chat archives for getredash/redash

22nd
Mar 2016
ChiragKParmar
@ChiragKParmar
Mar 22 2016 03:09
@arikfr is there any way of performing multiple (nested) filtering e.g. Filter on Employer -> Office Location (choices limited by what is being selected at employer)?
ChiragKParmar
@ChiragKParmar
Mar 22 2016 03:25
Also, I am getting "Error running query: Worker exited prematurely: signal 9 (SIGKILL)" on running simple select * query from a Redshift table (15 million rows), have you encountered this before?
Neil Shah
@shahneil88
Mar 22 2016 04:25
@mariuszlitwin - I have encountered that issue before. I feel its just redash needs more time. I used to logout and login again and could see data source/groups appearing there
Arik Fraimovich
@arikfr
Mar 22 2016 07:13
@mariuszlitwin I didn't encounter this before. Let me know if it happens again.
@vorakumar depends on your needs and type of setup, but generally I would only split out the database and leave Redis on the same machine.
@ChiragKParmar re. the SIGKILL -- most likely it's the OOM (out of memory) killer killing the worker. if you ran "select *" without any limits, it will try to bring all all the 15M records...
@ChiragKParmar nested filtering - not at the moment.
Mariusz Litwin
@mariuszlitwin
Mar 22 2016 07:17
@arikfr I'm currently reaching office. I have to revert my changes to Postgre setup and then, if you provide me with path to proper logs I can provide you with that. I had some troubles creating users to ("Internal Server Error") but never found proper logs.
Arik Fraimovich
@arikfr
Mar 22 2016 07:18
@mariuszlitwin the logs are at /opt/redash/logs/api_error.log
Mariusz Litwin
@mariuszlitwin
Mar 22 2016 07:49
Ok, so here is my user creation error: http://pastebin.com/H4MqEMaC
Arik Fraimovich
@arikfr
Mar 22 2016 07:50
@mariuszlitwin and this is still happening?
Mariusz Litwin
@mariuszlitwin
Mar 22 2016 07:56
System generated this error about 5 minutes ago
Arik Fraimovich
@arikfr
Mar 22 2016 07:57
@mariuszlitwin can you check if: 1. you have a group named default? 2. is its type 'builtin' or 'regular'?
Mariusz Litwin
@mariuszlitwin
Mar 22 2016 07:58
@arikfr There is 'default' group when I check this in front-end
Arik Fraimovich
@arikfr
Mar 22 2016 07:58
yes, but can you check in the db what value it has in the 'type' column?
Mariusz Litwin
@mariuszlitwin
Mar 22 2016 07:59
Give me a moment and I will try some command-line magic to determine it's type
@arikfr So 'default' group is regular type
Arik Fraimovich
@arikfr
Mar 22 2016 08:04
@mariuszlitwin this is the problem, update its type to builtin. I wonder if the migration we have in 0.10 messed this up.
Mariusz Litwin
@mariuszlitwin
Mar 22 2016 08:05
@arikfr Cool, thank you. I will report on results shortly.
@arikfr Ok, everything works fine. Thank you.
Arik Fraimovich
@arikfr
Mar 22 2016 08:07
thank you for reporting this!
ChiragKParmar
@ChiragKParmar
Mar 22 2016 13:56
@arikfr thanks. Table with 15 M record is already pre-aggregated table. What do you suggest in this case then?
Arik Fraimovich
@arikfr
Mar 22 2016 13:56
@ChiragKParmar what would you want to do in the UI w/ 15M records?
ChiragKParmar
@ChiragKParmar
Mar 22 2016 13:59
:) I don’t user to see all the records as table, but want to see trend over time for individual clients (that they can filter on)
Sorry for the confusion, I should have explained context first.
Arik Fraimovich
@arikfr
Mar 22 2016 13:59
then create a query that returns the trend data
as for filtering - currently there is limited support for parameters, which will be extended and allow to better implement filter scenarios
ChiragKParmar
@ChiragKParmar
Mar 22 2016 14:03
Table already contains trend data (preaggregated per day per client).
Arik Fraimovich
@arikfr
Mar 22 2016 14:04
each line is something like {day, client, value} ?
you can create a query that is filtered by some date range and client(s) to return only the relevant part of the data
ChiragKParmar
@ChiragKParmar
Mar 22 2016 14:05
hm..let me try with last 30 days. Thanks @arikfr. BTW I love redash :)
Arik Fraimovich
@arikfr
Mar 22 2016 14:05
:)
I suggest you run a count query first to make sure the amount of rows makes sense
just to avoid waiting long time for nothing
(gotta go now, will be available again later)
ChiragKParmar
@ChiragKParmar
Mar 22 2016 14:07
cool! Thanks @arikfr
Kumar Vora
@vorakumar
Mar 22 2016 15:43
@arikfr thanks! Our usage is going to be low. So, I like your suggestion. I will just use RDS for postgres and leave Redis on the same EC2 box.
Arik Fraimovich
@arikfr
Mar 22 2016 16:05
@vorakumar perfect setup.
ChiragKParmar
@ChiragKParmar
Mar 22 2016 16:10
@arikfr last 30 days didn’t work well (gave the same error), last 7 days for one client worked fine. Pre filtering for per client is not going to work as I will have to create query and dashboard for each client. Any suggestions?
Arik Fraimovich
@arikfr
Mar 22 2016 16:11
@ChiragKParmar use parameters - demo.redash.io/queries/146/source?p_action=create
then create another query that returns a row per client & link to parameter based dashboard/query
(how many clients there are?)
ChiragKParmar
@ChiragKParmar
Mar 22 2016 16:23
what do you mean by "create another query that returns a row per client”? I modified my query to accept client id as parameter.
Arik Fraimovich
@arikfr
Mar 22 2016 16:24
@ChiragKParmar something like this: http://demo.redash.io/queries/520#table
ChiragKParmar
@ChiragKParmar
Mar 22 2016 16:26
ahh! Very cool! Thanks @arikfr
eschultze
@eschultze
Mar 22 2016 18:26
is there an easy way from preventing user to change its password? I tried to redirect /users with nginx but I'm facing some issues. FYI - I have a generic read only user
Arik Fraimovich
@arikfr
Mar 22 2016 18:27
@eschultze not without changing code. but maybe you can use public dashboards now? getredash/redash#906
eschultze
@eschultze
Mar 22 2016 18:34
nice! I'm sorry, but where can I enable this feature?
Arik Fraimovich
@arikfr
Mar 22 2016 18:34
@eschultze you need to upgrade to the 0.10.0 RC version and then you will see a share button on your dashboard
eschultze
@eschultze
Mar 22 2016 18:39
@arikfr using the fab file it upgrades to 0.9.2.b1536. How can I do that?
Arik Fraimovich
@arikfr
Mar 22 2016 18:40
run it with deploy_latest_release:stable=false instead of deploy_latest_release.
eschultze
@eschultze
Mar 22 2016 18:40
thanks!
Arik Fraimovich
@arikfr
Mar 22 2016 18:40
note that you might need to fix your group types after upgrade
I will provide some details a bit later when I'm here again
eschultze
@eschultze
Mar 22 2016 18:40
ok