Hi,

I've set up the sentry client/server on a couple of sites and watched it to collect data for a couple of days. Maybe something in my setup is wrong, but the reports I'm getting are totally wrong and misleading. For example, the installed modules are not evaluated properly; screenshot attached. On this site, "Clean URLs", "Global Redirect", "Pathauto", "Nodewords", etc. are installed and enabled; however, Sentry reports: "Module is not enabled", and "Variable not set properly".

The "HTTP availability" report I do not understand. It looks like so:

Sun, 2011-12-18 13:27	20 ms
Sun, 2011-12-18 13:27	15015 ms
Sun, 2011-12-18 13:27	9462 ms
Sun, 2011-12-18 13:27	9037 ms
Sun, 2011-12-18 13:27	3028 ms
Sun, 2011-12-18 13:27	3040 ms
Sun, 2011-12-18 13:27	9052 ms
...

Opposed to that, the "Update period" is set to "1 hour". Why are there multiple checks/responses per second?

The "Cron" report looks identical:

Sun, 2011-12-18 13:27	20 ms
Sun, 2011-12-18 13:27	15015 ms
Sun, 2011-12-18 13:27	9462 ms
Sun, 2011-12-18 13:27	9037 ms
Sun, 2011-12-18 13:27	3028 ms
Sun, 2011-12-18 13:27	3040 ms
Sun, 2011-12-18 13:27	9052 ms
...

Again, multiple records per second, and actually the monitored site broke down, maybe because it ws flooded with such requests?

For now, I have disable the sentry client/server on all my sites until I figure out what sub-modules actually work and if sentry monitoring might have harmful side-effects. I'd really appreciate a statement about the state of this project on the project page - is sentry supposed to be production ready, in full or in (which) parts, or is it currently more intended as a proof of concept?

Thanks!

CommentFileSizeAuthor
sentry-wrong-report.jpeg240.79 KBasb

Comments

snufkin’s picture

First of all let me clarify the status of the module. The server module is under development, thus the dev status. There are quite a few companies who helped out with the development who are using it, so overall its more or less usable.

Regarding the http and cron reports - this report does indeed seems to be buggy, but since i dont have a setup at hand I can't check if i can get these results too. As for the other reports, these should report back correctly, I would check the status of the site if the keys match - admittedly some of the error handling is far from perfect.

asb’s picture

Thanks for the clarification. I could indeed trace back the dead site to "monitoring" requests which led to a similar effect like a DoS attack, so I'd be very carefull when testing this module.

snufkin’s picture

Title: Sentry reports wrong status » cron/http check triggers way to many times per run

Well that should not happen though, this is the first time i hear of this. The HTTP request should only trigger once per cron run. I changed the title to reflect this.