Hi Epic,
We’re currently testing Horde out to perform Packaged Build using a vanilla BuildAndTestProject.xml build graph (the one provided in Engine/Build/Graph/Tasks without modification). It works fine, except we sometimes get random HttpTimeout when artifacts get uploaded at the end of some compilation steps. We’re having trouble finding out the cause of these timeouts as they happen randomly between jobs, CLs, agents, time of the day, etc. See attached file for example of a failing job.
Similar topics here seem to solve this issue by simply getting more firepower (upgrading network/bandwidth, changing ISP, etc), or hardcoding a higher timeout in Horde source code. We cannot/do not want do that so we’re trying to pinpoint the cause of these timeouts.
It feels like a network/hardware issue, but it could come from the agent, server, storage, network between these machines, firewall, or a combinaison of those.
Agents are running on independant machines, Horde server on a Linux VM, and Storage is hosted on intenal hardware (i.e. no AWS). Artifact uploads seems to have to go through the Horde server to end up on Storage.
Monitorings of the server, the agents and the storage do not indicate obvious issue (we do not see 100% usage of CPU, Disk or RAM usage for example).
Something that could help us understanding what is going on would be to be able to trigger artifact upload manually with a dummy of arbitrary size, at arbitrary time, repeatedely and consistently, instead of having to spawn jobs and wait for a whole compilation step to run.
As we’re still kind of new to Horde as a Build Sytem, what would we an easy way for us to create a dummy artifact of several GB and just perform its upload to Horde to trigger these timeouts?
Thank you for your help.
Kind regards
This question was created in reference to: [Horde Http timeout when copying artifact from agent -> [Content removed]
[Attachment Removed]