Content_CheckURL rule interfering?
analogger
i'n general, i've been really impressed with serverscheck. but one problem area that i have applied it to has me concerned. i don't have enough data to be certain---and i can't reproduce the problem in order to test it properly. but i suspect that some interaction between the monitoring rule 'Content_CheckURL' and the server that it was monitoring interfered with the return of this url to availability from a down state.
i was monitoring a https:// url that displays a log on form for a web-based (ASP) database application. for an as yet unknown reason, this url becomes unavailable to our site for periods ranging from a few minutes to a few hours.
serverscheck was doing a good job of logging down time and i was using other rules to show that the site could be ping'ed even if the page was not available...
over the course of several days serverscheck did monitor and log several failures and spontaneous recoveries of this page, but then the server failed to spontaneously recover for
an unprecedented period of over 24hours. when i stopped the monitoring, however, the server recovered spontaneously within a few minutes (as empirically tested via browser). something like this, but less obvious to me at the time, happened once before.
i don't know why this page fails--i don't have access to the server. the http:// version of the page will display when the https:// page is failing. and it only fails from our site, not from elsewhere.
so i am wondering if anyone else has seen this happen? i don't know exactly how the Content_CheckURL works, but i am wondering if it is possible that this secure url is initially failing because of too much browser activity from our IP address or something, and then the monitoring rule is somehow keeping our site shut out once it is down, on account of the monitoring activity itself?
perhaps there is some way that i could test this behavior by setting up a set of monitoring rules.... how else could i test this page without fear of affecting the problem? would the status/headers/url_exists tests be any less likely to cause this problem?
i was monitoring a https:// url that displays a log on form for a web-based (ASP) database application. for an as yet unknown reason, this url becomes unavailable to our site for periods ranging from a few minutes to a few hours.
serverscheck was doing a good job of logging down time and i was using other rules to show that the site could be ping'ed even if the page was not available...
over the course of several days serverscheck did monitor and log several failures and spontaneous recoveries of this page, but then the server failed to spontaneously recover for
an unprecedented period of over 24hours. when i stopped the monitoring, however, the server recovered spontaneously within a few minutes (as empirically tested via browser). something like this, but less obvious to me at the time, happened once before.
i don't know why this page fails--i don't have access to the server. the http:// version of the page will display when the https:// page is failing. and it only fails from our site, not from elsewhere.
so i am wondering if anyone else has seen this happen? i don't know exactly how the Content_CheckURL works, but i am wondering if it is possible that this secure url is initially failing because of too much browser activity from our IP address or something, and then the monitoring rule is somehow keeping our site shut out once it is down, on account of the monitoring activity itself?
perhaps there is some way that i could test this behavior by setting up a set of monitoring rules.... how else could i test this page without fear of affecting the problem? would the status/headers/url_exists tests be any less likely to cause this problem?
This discussion has been closed.
Comments
Activity is being logged in a log file located in the logging subdirectory of ServersCheck.
The first step is to assess the error that ServersCheck received and that caused the rule to go from OK to DOWN.
The next step is to verify the log files of the web server running the web site and to see if errors can be detected there (500 errors are logged in the log file).
The ContentCheck performs a GET request to the specified URL and loads the complete HTML file into memory. It then scans the page to see if the specified string can be found. The check uses a custom HTTP_USER_AGENT.
unfortunately i don't operate the server that i'm monitoring--so i don't have access to the logs. i'd have probably solved the problem long ago if i had. they are reluctant to believe that it's their problem, which is why i've been trying to collect some empirical data.
the servercheck logs show that the url check is humming along reporting http code 200 and consistent download times. then there's no code and no time, and after several tries there's the annotation " A connection with the server could not be established." i was running a simultaneous ping and dns check and they did not fail.
but the real gist of my question remains: have you seen situations before where monitoring a url when it is down has prevented it from coming back up again?
A PING request is not the same as a HTTP GET request as this happens on a different layer of the server. The PING request is on a OS/network level; the HTTP GET is on an application layer.
Depending on the infrastructure, some webservers/firewalls are configured to block similar requests happening to frequently on a webserver (this is to prevent Denial Of Service attacks - DoS)
thanks for your reply. the ability to monitor, compare, and graph these unrelated functions simultaneously is what makes this such a powerful troubleshooting tool. in this case i can demonstrate that my internet connection is working, the server is up, but this application is down... i've tried several one-trick ponies that only monitor PING or DNS or whatever. this is super.
is there an API for interfacing with one's own physical sensors?
Regards,
Forum Administrator