ambari-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jonathan Hurley <jhur...@hortonworks.com>
Subject Re: Review Request 43748: Add alerts for HAWQ components status
Date Fri, 19 Feb 2016 02:25:15 GMT


> On Feb. 18, 2016, 9:05 p.m., jun aoki wrote:
> > ambari-server/src/main/resources/common-services/HAWQ/2.0.0/alerts.json, line 24
> > <https://reviews.apache.org/r/43748/diff/1/?file=1258737#file1258737line24>
> >
> >     Is type OK for checking a process is down?
> >     I don't know if a type called "PROCESS", but otherwise maybe "SCRIPT" to issue
same ps with process names?
> 
> bhuvnesh chaudhary wrote:
>     Good point. I see a possibility of some other process occupying the port and might
be listening on it, even though Hawq process may be down. In that case, an alert will not
be sent, but such occurence may be limited.
>     
>     Also, as a standard/recommended practice we should have hawq being run on reserved
port, since if we let HAWQ run on the current unreserved port range, HAWQ component startup
can also fail if some process is already occupying that port. That change should also resolve
the issue mentioned above.
>     
>     Ensuring check for process name is something which we can possibily have an enhancement
on alerts and add a field regex_pattern in alerts.json to search for the process on the port
to avoid such issue, instead of creating a separate script for process check since type PORT
is common across all the services and used similarly. The same issue can manifests for all
the services.
>     
>     If you are okay, i will create a separate JIRA on alerts for enhancement and drop
this issue here ?
>     and for ensuring that HAWQ is run on a reserved port range will work internally ?

Typically, yes, using a PORT to check and see if a process is down is good enough. Typically,
if there's a WEB endpoint we'll use that. In cases where we also want to test functionality
of the component, a SCRIPT alert will be written.


- Jonathan


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/43748/#review119778
-----------------------------------------------------------


On Feb. 18, 2016, 7:59 p.m., bhuvnesh chaudhary wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/43748/
> -----------------------------------------------------------
> 
> (Updated Feb. 18, 2016, 7:59 p.m.)
> 
> 
> Review request for Ambari, Alejandro Fernandez, Jaimin Jetly, jun aoki, Jonathan Hurley,
Jayush Luniya, Oleksandr Diachenko, Richard Zang, Sumit Mohanty, and Yusaku Sako.
> 
> 
> Bugs: AMBARI-15105
>     https://issues.apache.org/jira/browse/AMBARI-15105
> 
> 
> Repository: ambari
> 
> 
> Description
> -------
> 
> This patch introduces alerts for HAWQ components based on PORT. The below components
are covered.
> - Master
> - Standby
> - Segment
> 
> 
> Diffs
> -----
> 
>   ambari-server/src/main/resources/common-services/HAWQ/2.0.0/alerts.json 3119a0c 
> 
> Diff: https://reviews.apache.org/r/43748/diff/
> 
> 
> Testing
> -------
> 
> yes, manual.
> 
> 
> Thanks,
> 
> bhuvnesh chaudhary
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message