Hi,
I'm storing artifacts within Go (2.2 Enterprise) to use in subsequent pipelines.
Fetching the artifact is taking what appears to me to be far too long (It is a largish tar file of approx 600Mb), but the agent is installed on the same Red Hat server as the Go server so I'm thinking that fetch is actually just copying the file from one location on the server to another. If I do this by hand (using the cp command) it takes a few seconds, but using Fetch within Go it takes over 10 minutes.
Any thoughts as to whether I have something wrongly configured? or an explanationas to what Fetch is doing that causes the length of time to be so long.
Steve
Comments
2 comments
Hi Steve,
Eventhough the agent is installed on the same machine as the server, the fetch operation is not really a copy from one location to the other.
The agent communicates with the server via https and downloads the artifact from the server via https. Hence, that may take some time in case of a large file. There are a few reasons due to which the download maybe slower. Could you answer a couple of questions which might give us an idea as to the long time being taken in the download ?
1. Are the agent and server installed on different disks?
2. Is the server very busy with other operations like, pipelines getting triggered, config changes being made, artifacts downloads and uploads to and from the agents?
Regards
Shilpa
Hi Shilpa,
Answers to your questions:
1. No agent and server are installed on the same disk (indeed I have 4 agents installed on this disk).
2. The server is a powerful server 8cpus, 16Gb Ram. So whilst there are all the operations you describe in process the server itself is always under utilised.
Based on your comment I'm wondering if having a separate disk for the agents may prove to be more efficent.
Steve
Please sign in to leave a comment.