shinken-monitoring/mod-webui

Downtime: Can't remove downtime on service with a 'slash' in the service name

Closed this issue · 27 comments

serge-marie commented

Hi,

I can't delete downtime on several services, I think that comes from the slash in the name.

Example :
LINUX-DSK_/

I have a "Generic modal" windows" :

image

Thank's in advance

Serge

maethor commented

I will look at it.

maethor commented

Indeed, I just saw the source code, in shinken-actions.js we split the element name on the /, because it is supposed to be the element of separation between the host name and the service name.

Right now I can't think about a nice fix, because if we admit / are allowed in hosts or services names, the / could be everywhere.

I think you should really avoid having / in service names.

I'm curious @serge-marie, is it the only thing that don't work in the webui? Everything else is OK with the /?

serge-marie commented
mohierf commented

@maethor : a solution would be to use another separator in the HTML built with the templates. Let's say host_name$service_description and then split on $ rather than /

In helper.py:

    def get_uri_name(self, elt):
        name = elt.get_name()
        if 'service_description' in elt:
            name = "%s$%s" % (name, elt.get_name())
        return name.replace(' ', '%20')

This function is supposed to be called for each parameter from Python to Javascript ... I hope so :)

maethor commented

I suppose it is used almost everywhere, but not in shinken-actions.js get_elements function.

And what if some users have services with $ in them?

mohierf commented

As far as I remember, $ is not allowed in a service description.

See illegal_object_name_chars that defines the non-authorized characters in objects names is as default: ``~!$%^&*"|'<>?,()=`

maethor commented

oh, that's nice. So | may be ok too…

I will see if I can replace it in all the app.

maethor commented

It seems like after some macros and replacements, the command sent by the WebUI to shinken uses ; to separate the host and the service:

4807:[1496236089] INFO: [broker-master] [WebUI-actions] got command: SCHEDULE_SVC_DOWNTIME with args: ['www.sysnove.fr', 'Http', '1496236080', '1496408880', '1', '0', '0', 'admin', 'Downtime scheduled from WebUI by admin'].
4808:[1496236089] INFO: [broker-master] [WebUI-actions] external command: [1496236089] SCHEDULE_SVC_DOWNTIME;www.sysnove.fr;Http;1496236080;1496408880;1;0;0;admin;Downtime scheduled from WebUI by admin.

So I don't understand why ; is the only character not listed in your list @mohierf 😄

Yet I'm not sure to understand how it how it works, because I don't see where that ; is defined. But I will find it.

So ; should be a pretty good character to use, since it's what shinken is using?

mohierf commented

Beware @maethor that the ; is the Nagios legacy separator for the external commands. It is not forbidden to use it but it is a Nagios trick 😉 Never use this character in a service_description or forget the external commands ...

See action.py, line 100 where an ExternalCommand is built to be push to the broker that runs the WebUI module ...

maethor commented

OK, I see it no, in the try catch.

So, that's exactly what I'm saying, ; is already forbidden in host_name and service_description, so I can use it as a separator instead of /.

Anyway, if you think $ is the better choice, let's use it 😄

maethor commented

What I find strange is that you succeed to put a downtime, but not to remove it. I will try to reproduce this bug tonight.

maethor commented

2017-06-01-210409_603x293_scrot

So it doesn't seem to come from the /. Everything is working fine on my test : ack, downtime, remove downtime, remove ack.

  • @serge-marie can you precise exactly how to reproduce your problem?
  • Is it the only thing that don't work in the webui?
  • Everything else is OK with the /?
  • What is the exact version of your webui?

You can answer in french if that's easier for you.

maethor commented

@mohierf OK, so now I think I understand the mecanics between shinken-actions.js and action.py. It's really ugly, I mean… it's awful… but I don't see any bug in it. / in service names seems to be managed properly.

  • The first / is kept, but the nexts are replaced by $SLASH$ in shinken-actions.js/get_elements function
  • Then in action.py, the / is replaced by ;, and then $SLASH$ are replaced by /.

I could rewrite that, but if it's working…

I have a question. Do you this / should be also possible in host_name? I really don't think shinken core can manage that.

serge-marie commented
maethor commented

@serge-marie if you have the possibility to upgrade to develop, I think you should try to reproduce the bug on it. If not, please answer my questions, yes :)

Good night ;)

serge-marie commented
serge-marie commented
maethor commented

I'm sorry but your image doesn't display. Could you please respond directly on github?

serge-marie commented

Hi,

This is my webui version :

0

This is how I can reproduce the problem :

I have an host :

1

I use "Select all elements" -> "Execute an action" -> "Schedule a downtime".

I have :

2

Services are "downtimed" :

3

I remove the downtime with "Click on hostname" -> "Delete all downtimes"

I have :

4

And when I refresh the page, downtimes on services are always set :

5

Only the downtime on the host is removed :

6

If I remove downtime on each service with the button "7" at the right of the service.

When I do it on a service without a "/" in the name, it's working :

9

And the downtime disappears.

But on a service with a "/" in the name (the first one only) I have the same window but the downtime is still here after click.

After that, on another service with a "/" I have also : (on the same service too)

10

After all the click :

11

All downtimes on those services are still presents ...

12

I hope that is helpfull for you to understand my problem.

Serge

serge-marie commented

I forgot to precise that we use mongodb-dt-ct-retention-scheduler module.

For data retention of comments and downtimes.

maethor commented

Ok, so you are still on 2.4.2c and you describes 2 differents problems that I I have found and fixed in 2.5:

  1. When we add a downtime on a host, by default the downtime is added on all the services. But when you remove the downtime, it is removed only on the host. This is fixed.
  2. The problem with the / in the service name. This is fixed.
serge-marie commented

OK thank's

How can I update it easily ?

Because on SHINKEN-IO :

image

Serge

maethor commented

2.5 is not published yet. I still have a major regression to fix and then I will publish it. I can't give you a deadline, could be a week or two.

serge-marie commented
maethor commented

@serge-marie As promised, I published 2.5.0 today. It should be available tomorrow on shinken.io. Please close this issue if the new version is fixing your problem.

serge-marie commented
serge-marie commented