Reset Search



KB43699 - ZCLI output "HTML Error: Failed to load page" occurs intermittently when running concurrent SOAP scripts

« Go Back


Last Modified Date5/25/2018 9:08 AM
ZCLI output "HTML Error: Failed to load page"  occurs intermittently when running concurrent SOAP scripts using CRON. The following is logged by the Traffic Manager when the intermittent error occurs:

Admin Access logs: - admin [13/Feb/2018:13:00:02 -0800] "POST /soap HTTP/1.1" 502 2317 

Admin Error logs:
[13/Feb/2018:13:00:02 -0800] WARN admin:FCGI: Read error: Connection reset by peer 

What is the source of these intermittent failures?
Problem or Goal
The source of the intermittent 502 error is a purposeful change introduced in version 10.2 to restart the SOAP process at the 1000th request to avoid the underlying PERL process from a memory growth leak.  This memory leak as a result of  several MB, several hundred SOAP requests, and eventually several 10000 requests would eventually cause the machine to run out of memory.

When a concurrent SOAP request is made while the SOAP API is restarted, then this results in a 502 bad gateway message.  This is documented in the 10.2 Release Notes:

Fixed a problem where the SOAP FCGI process memory usage could grow indefinitely and cause an outage. If soap!max_requests is set the process will restart after that many requests or 1000 otherwise. Clients that make frequent requests of the SOAP API may see a 502 bad gateway, and should re-send the request if encountered.
The solution is to resend the request  in the event that 502 error is encountered.

A better alternative would be to run the SOAP requests sequentially via a single CRON script instead of running concurrent SOAP requests.   
Related Links
Attachment 1 
Created ByRichard (Rick) Henderson



Was this article helpful?



Please tell us how we can make this article more useful.

Characters Remaining: 255