Reduce the default oVirt open timeout to 1 minute #15099
Conversation
Currently the timeout used when opening connections to oVirt isn't explicitly set. That means that when trying to connect to an oVirt system that doesn't exist, or that doesn't respond, it may take as long as the TCP timeout till the error is detected. In a typical setup on top of Linux this timeout is approx 2 minutes and 8 seconds. But in production environments the web server that is in front of the ManageIQ user interface has a request timeout explicitly set to 2 minutes. This means that validation of credentials for oVirt systems that don't exist, or don't respond are ignored by the ManageIQ user interface, because it gives up before the error is detected by the ManageIQ server. To avoid that issue this patch changes the default connection timeout used by the oVirt provider to 1 minute. https://bugzilla.redhat.com/1448065
|
Checked commit https://github.com/jhernand/manageiq/commit/ec8664221bd59a129e63ea12092f44a8f41f9106 with ruby 2.2.6, rubocop 0.47.1, and haml-lint 0.20.0 |
|
@jhernand Maybe you should move the whole ems_redhat section of settings to https://github.com/ManageIQ/manageiq-providers-ovirt/blob/master/config/settings.yml |
|
@chessbyte yes, that makes sense. How are upgrades managed in that case? I mean, what will happen for users that already did changes to their global settings file? Will the settings in the provider specific file override the settings in the global file or the other way around? |
|
@jhernand Users shouldn't be changing anything in For extra credit, it would be nice if the key was also changed to |
|
@jhernand Please add a Fine label |
|
@miq-bot add_label fine/yes |
…eout Reduce the default oVirt open timeout to 1 minute (cherry picked from commit 7e05340) https://bugzilla.redhat.com/show_bug.cgi?id=1461860
|
Fine backport details: |
Currently the timeout used when opening connections to oVirt isn't
explicitly set. That means that when trying to connect to an oVirt
system that doesn't exist, or that doesn't respond, it may take as long
as the TCP timeout till the error is detected. In a typical setup on top
of Linux this timeout is approx 2 minutes and 8 seconds. But in
production environments the web server that is in front of the ManageIQ
user interface has a request timeout explicitly set to 2 minutes. This
means that validation of credentials for oVirt systems that don't exist,
or don't respond are ignored by the ManageIQ user interface, because it
gives up before the error is detected by the ManageIQ server. To avoid
that issue this patch changes the default connection timeout used by the
oVirt provider to 1 minute.
https://bugzilla.redhat.com/1448065