active agents do SUCK on HM restart

rasc · Post by **rasc** » Mon Oct 12, 2009 12:04 pm

Folks,

you're not the only company where I do ask myself if you EVER used your own products.
We got 8 active agents (Enterprise license). And EACH time HM is (re)started EACH (active) agent produces for EACH test an error message (mail)!

Have you ever used your product? How can you have such a configuration?!

Same with NOT giving the choice if "unknown" is to be "ignored" OR "bad".

Would you mind to improve your product to a usable level?

Thank you very much.

P.S. I just restarted the HM machine and got 127 WRONG error mails. I might be a tiny bit pissed, so excuse my tone.

KS-Soft · Post by **KS-Soft** » Mon Oct 12, 2009 12:49 pm

1) We do not use HostMonitor for production monitoring. We are software developing company, not a Internet Service Providers. We have nothing to monitor except 1 web site (yes HostMonitor checks our site as well but I cannot call this "production monitoring").
However we are runnign several instances of HostMonitor 24/7 for testing purposes.

2) There is option to "ignore" Unknown test status. Just unmark "Treat Unknown as Bad" option located in Test Properties dialog

3) We always recommend different solution - setup 1 auxiliary test for each agent to check connection to the agent. Then use this test as Master test for all test items performed by the agent. Please read the manual or visit our web site for more information: http://www.ks-soft.net/hostmon.eng/mfra ... htm#Master
Manual in MS Word format included in the package (manual.doc file)

4) Also we recommend to use Active RMA when you cannot use HostMonitor itself or Passive RMA

Regards
Alex

rasc · Post by **rasc** » Mon Oct 12, 2009 1:07 pm

Alex,

1)

2) NO, there's NOT. in MOST tests. MOST tests do NOT have an option "Treat Unknown as Bad" located in Test Properties dialog.
Actually I just searched for one that DOES. Ping doesn't. LDAP doesn't. Traffic Monitor doesn't.
Or just show me (screen shot) where that option is. I was glad and would beg for pardon.

3) There is no "master test" for an active agent (despite for instance to ping a passive agent). Active agent is meant to be behind a firewall which does not (necessarily have to) respond to PINGs or other requests.

4?) ??? have you re-read your post??? Where's the relation to what I asked???

Thank you.

KS-Soft · Post by **KS-Soft** » Mon Oct 12, 2009 1:35 pm

NO, there's NOT. in MOST tests. MOST tests do NOT have an option "Treat Unknown as Bad" located in Test Properties dialog.
Actually I just searched for one that DOES. Ping doesn't. LDAP doesn't. Traffic Monitor doesn't.
Or just show me (screen shot) where that option is. I was glad and would beg for pardon.

Sure
http://www.ks-soft.net/images/hostmon7/tedit2.gif
If you do not see this option for some test then you don't see it for any other tests because some options could be "collapsed". Just click [+] button above "Help" button to "expand" Test Properties dialog.

There is no "master test" for an active agent (despite for instance to ping a passive agent). Active agent is meant to be behind a firewall which does not (necessarily have to) respond to PINGs or other requests.

There is no master test? Of course there is no such test (yet). You should create tests and alerts by yourself.
I do not see any problem here. Active RMA can easily ping localhost (127.0.0.1). If agent is connected to HostMonitor, HostMonitor will set "Host is alive" test status and perform dependant tests. If agent is not connected, HostMonitor will set "Unknown" status and will not start dependant tests until agent connects to HostMonitor.

have you re-read your post??? Where's the relation to what I asked???

Sure, I have read your post and my answer.
Answers #1, #2 and #3 respond to your questions. Item #4 is just a recomendation.
Actually items #2 and #3 are recommendations as well. If you want us to answer just a questions and do not provide any advises, no problem. Then you may ignore items #2, 3 and 4. I will answer specific questions only.

Regards
Alex

rasc · Post by **rasc** » Mon Oct 12, 2009 1:59 pm

Dear Alex,

in matters of 2)
I beg -as promised- (partially) your pardon.
I did not notice that tiny [+], my fault. Just, why don't you always show these options by default? The times of 600x480 VGA monitors are 15 years ago! We've got enough space on monitors today to display options in dialogs.

regarding 3):
a) it is NOT my job to define (with huge effort [e.g. port opening, etc] a master test for an active agent (aa)
b) a master test would NOT HELP when starting HM as HM would be able to connect whereas AA would not have (yet) connected back and the error was trown anyway [except 2) was configured].

Regarding 4) with "have you re-read your post??? Where's the relation to what I asked??? " I was referring to your 4) which has NO (recognizable to me) relation to my original post. Your 4) recommends the use of AA. I agree. But only if the issues stated in my original post were resolved (maybe a 10 or 15 minutes delay for AA before giving even an 'UNKNOWN' status)

Thank you

rasc · Post by **rasc** » Mon Oct 12, 2009 2:06 pm

oh, just a thought:

A GLOBAL setting (in Options dialogue) was very preferable.
Treat ALL unknown stati as BAD // Treat AA's unknown stati as BAD // Treat PA's unknown stati as BAD
etc....

Thank you

P.S. NOT to treat it as bad was like UNcheking your box in all dialogues

KS-Soft · Post by **KS-Soft** » Mon Oct 12, 2009 3:04 pm

a) it is NOT my job to define (with huge effort [e.g. port opening, etc] a master test for an active agent (aa)
b) a master test would NOT HELP when starting HM as HM would be able to connect whereas AA would not have (yet) connected back and the error was trown anyway [except 2) was configured].

a1) Well, if its not your job to setup tests and configure HostMonitor, may I speak with somebody who is responsible for this job?
a2) Sorry, I do not understand the problem. Huge effort?? You already have Active RMA installed on some remote system. Right? You already have some test items for this agent. Right?
Now you need just 30 seconds to setup 1 additional ping test using this agent. E.g.
- click Add button
- select agent from drop down list ("Test by" property)
- set Ping test method ("Test method" property)
- type 127.0.0.1 in "Address" field
- optionally change default test name and assign alert profile
- click Ok button
Then you will need several minutes to select all test items performed by this agent, click "Edit" button, add master test and click Ok button

b) a master test would NOT HELP when starting HM as HM would be able to connect whereas AA would not have (yet) connected back and the error was trown anyway [except 2) was configured].

You definately do not understand me

I am not telling you to setup HostMonitor to ping system where RMA is running directly. I am trying to tell you that you may setup Master test that should be performed by the same agent!!

Regarding 4) with "have you re-read your post??? Where's the relation to what I asked??? " I was referring to your 4) which has NO (recognizable to me) relation to my original post. Your 4) recommends the use of AA. I agree.

Not exactly. I am trying to say the following: use Active RMA only in case you cannot use HostMonitor directly or Passive RMA. Do not use Active RMA if you can use HostMonitor. Do not use Active RMA if you can use Passive RMA.
Sorry, if I am not clear

Regards
Alex

rasc · Post by **rasc** » Mon Oct 12, 2009 3:23 pm

Alex,

I am so sorry you feel agressed!
a1) I AM responsible to setup USEFUL tests.
a2) You might define 20 'workarounds', the ONLY REAL solution was "no feedback from AA-tests within x minutes after HM startup = no problem"!!!

I DO understand you, just you do NOT understand your custumer! Just give me the option to delay ANY error from AA until X minutes after HM startup!!!
I am not supposed to setup a "stop other tests"-test. I want a ready-to-use-product for the money you take.

Regarding 4)
Hmmm. It thought it was my choice. Not the manufacturers.

Thank you.

KS-Soft · Post by **KS-Soft** » Mon Oct 12, 2009 4:11 pm

There are many options... e.g. you may setup Master test, you may disable "Treat Unknown status as Bad" option, you may use "Startup delay" option, you may change alert settings...

However our software is not "out of the box" solution. It should be configured by network administrator.
If you do not want to manage software, if you think its not suitable for you, I am sorry. You may send request to sales@ks-soft.net, provide your order number. If you ordered software recently, we can cancel your order and return money to your account.

Regarding 4)
Hmmm. It thought it was my choice. Not the manufacturers.

Sure, everything is your choice. Including software you want to use. If you use our software and you ask us for help, I assume you want to hear my answer.
Probably I was wrong. Looks like you do not ask for any advice, you just want to show how you angry. Ok, no problem. I got the picture. We can return your money so you can choose different software.

Regards
Alex

KS-Soft · Post by **KS-Soft** » Mon Oct 12, 2009 6:11 pm

Just give me the option to delay ANY error from AA until X minutes after HM startup!!!

1) There is "Startup delay" option located on Service page in the Options dialog

2) Probably you are right and we can add some additional option specifically for Active agents. May be in next version...

3) "Just give me the option!!!" sounds a little strange in this case. You can demand to fix some critical problems. But we do not have to implement new options when there is easy workaround or similar option. Pretty often we add new options requested by our customers however this does not mean we have to do this immediately.

Regards
Alex

rasc · Post by **rasc** » Mon Oct 12, 2009 10:39 pm

I think your sarcasm and finickiness is misplaced. I won't answer that.

rasc · Post by **rasc** » Tue Oct 13, 2009 12:19 am

KS-Soft wrote: 1) There is "Startup delay" option located on Service page in the Options dialog

That option is not applied if HM is started as application (not service). I just re-checked. Set Startup delay = 5min. Start HM. First error mail after 5 seconds.

ldean · Post by **ldean** » Mon Oct 19, 2009 12:09 pm

dear rasc,
please stop being such a boner.

PS - why don't you just disable the alerts before you restart HM, then re-enable them afterwards?

KS-Soft · Post by **KS-Soft** » Mon Oct 19, 2009 12:19 pm

We already added new option in version 8.28 (should be available soon

Regards
Alex

jivetolkein · Post by **jivetolkein** » Tue Nov 24, 2009 7:31 am

The mails from restarts of active RMAs are easily avoided by applying some intelligence to the processing of alerts - we use a multi stage check (it cycles through checking again more quickly than scheduled) that I adopted from Nagios, so there is a 'soft' error period before it's established to really be a problem. Else any kind of network glitch can cause lots of false positives. This and parent/child test relationships can minimise these issues.

We use HM to monitor over 500 nodes across a fair chunk of the world. It isn't perfect but I've yet to find anything that couldn't be fixed by configuration changes, scripting or asking the authors here. Your tone is unhelpful and belligerent and if these kind of things get you this angry I suggest you hire a consultant who understands monitoring and your requirements and make it their job to correct - then pay the sizable bill you'll be getting. Host Monitor is a walk in the park compared with some of the heavy weight offerings I also have to manage (Tivoli, OpenView etc.), and I think you need to get a grip on the fact that effective monitoring is an adaptive process and might require some ongoing work from yourself to keep it in trim.