adoptium
Correct ARMv8 to aarch64 in Jenkins
Found as a part of https://github.com/AdoptOpenJDK/openjdk-infrastructure/issues/1716#issuecomment-755202080
As suggested by @sxa (https://github.com/AdoptOpenJDK/openjdk-infrastructure/pull/1806#pullrequestreview-562539987), we should change the Jenkins name of the following machines, so they match the inventory.yml, i.e. remove the l from armv(7|8)l:
- [ ] build-packet-ubuntu1804-armv8l-1
- [ ] docker-scaleway-ubuntu1604-armv7l-1
Should just be a case of changing their names, and restarting the Jenkins agent.
Clearing assignment as I don't have access to the machines to restart the Jenkins agent
I'm tempted to switch them all over to aarch64 instead of armv8 since I think that name is clearer to most people, and it's consistent with uname and what we use on the download site ... And the build job names
Yeah, that seems fair - should I rename the issue ? Should ARMv7 remain ARMv7? (I noticed that the job just refers to it as arm)
@sxa I'm happy to pick this up, as it's blocking me on #1716 , but presumably I'm going to need access to all of the arm64 nodes, so I'm able to restart the Jenkins Agents.
(The following is a list of machines that I don't have access to. I do have it for the rest of the ARMv8 nodes in the inventory)
| Node in Inventory | IP | Other Info |
|---|---|---|
| build-linaro-centos76-armv8-2 | 213.146.141.123 | |
| build-packet-ubuntu1804-armv8-1 | 139.178.82.234 | |
| docker-packet-ubuntu1604-armv8-1 | 147.75.77.146 | |
| docker-scaleway-ubuntu1604-armv7-1 | 51.158.73.136 | |
| test-macstadium-macos11-arm64-1 | 199.7.163.51 | (Blocked by #1855 ) |
Quick update on this - the Linaro machines are not currently in production, the scaleway ones are going to get decomissioned (and are armv7 not armv8), the packet aarch64 ones listed above are also to be decomissioned soon. We'll need to consider what to do with the macos ones ...
Still considering keeping armv8 since aarch64 will become ambiguous across armv8, armv8.2 (Required by Dragonwell) and the upcoming armv9
After some discussions with someone at ARM I've come to the conclusion that standardising on aarch64 is the right approach for the naming, so yes we should go for that :-)