Rolled out improvements have been verified to address the issue. * New Deployments UI with tabbed design * Deployments and Inventory database connection handling improvements * Deployments API performance improvements
Posted Aug 27, 2018 - 12:14 UTC
A fix has been implemented and we are monitoring the results.
Posted Aug 23, 2018 - 13:09 UTC
HTTP 504 errors are caused by service taking too long to reply in high traffic spikes. They used to be masked as HTTP 502 on client side and 499 on our side by load balancer timeouts that have been increased before the weekend. This allows us to distinguish disconnecting clients from internal timeouts however also changed error code on the client side.
We are evaluating options to improve response times in during those periods.
Posted Jul 31, 2018 - 08:13 UTC
Users experienced sporadic 504 errors returned by Deployments API though the weekend. The root cause for this is still being investigated however we have increased the overall capacity today as it is likely load-related.