Hi,
We are encountering a very strange behaviour in web check scenarios for some hosts. Parsing the page fails randomly, but we are unable to reproduce it with many other different tools like wget or curl... and it only happens when the required string is set to </html> (to ensure the page is rendered completely).
I've patched the zabbix server based on the code proposed in the comments of bug ZBXNEXT-101 to log the contents of the failing web page... and there, the last byte is missing, thus resulting in a "Page didn't match" error... but when checking a tcpdump file, the page gets transferred completely, it seems to get cut off somewhere inside the http poller code... but just for some requests of the corresponding URL, sometimes it works, sometimes it doesn't
Anyone encountered this problem as well?
Cheers,
Matthias
We are encountering a very strange behaviour in web check scenarios for some hosts. Parsing the page fails randomly, but we are unable to reproduce it with many other different tools like wget or curl... and it only happens when the required string is set to </html> (to ensure the page is rendered completely).
I've patched the zabbix server based on the code proposed in the comments of bug ZBXNEXT-101 to log the contents of the failing web page... and there, the last byte is missing, thus resulting in a "Page didn't match" error... but when checking a tcpdump file, the page gets transferred completely, it seems to get cut off somewhere inside the http poller code... but just for some requests of the corresponding URL, sometimes it works, sometimes it doesn't
Anyone encountered this problem as well?
Cheers,
Matthias