| Summary: | autobuild hasn't been run since July 2018 | ||
|---|---|---|---|
| Product: | Infrastructure | Reporter: | David Walser <luigiwalser> |
| Component: | Others | Assignee: | Sysadmin Team <sysadmin-bugs> |
| Status: | RESOLVED OLD | QA Contact: | |
| Severity: | normal | ||
| Priority: | Normal | CC: | pterjan, sysadmin-bugs |
| Version: | unspecified | ||
| Target Milestone: | --- | ||
| Hardware: | All | ||
| OS: | Linux | ||
| URL: | http://pkgsubmit.mageia.org/autobuild/ | ||
| Whiteboard: | |||
| Source RPM: | CVE: | ||
| Status comment: | |||
|
Description
David Walser
2019-01-15 23:03:50 CET
David Walser
2019-01-15 23:03:58 CET
CC:
(none) =>
pterjan I had been paying for it myself ($10 to $25 a month) to run on Google cloud but that was a bit expensive. I was in touch in June last year with packet.net which mentioned they are sponsoring some build machines to various projects but I ended up not having time to work on getting things setup there so didn't followup :( I was running it on a 32 core VM with 100G ram for tmpfs for about 20h once a week. A machine like a Scaleway C2L (€23/month, 8 core, 32GB ram, 250GB SSD) would probably allow to do 2 builds a week. Another alternative would be to not do full rebuilds of snapshots but extend the build system to do it in background. That would probably mean: - Have lower priority queue of packages, only to be touched when there is spare capacity (maybe even support for killing those builds if the normal queue grows too much) - Support discarding built packages - Have a way to identify those builds to be able to have history/logs for a package and publish a report But I don't think I would have time to work on such thing anytime soon. A temporary solution could be to do a few runs on the Scaleway machine until the release. Yeah please do a few runs until the release. I'm sure Mageia could spare some funds if need be, or I could if it comes to that. Pascal has been doing autobuilds when we need them. Closing. Status:
NEW =>
RESOLVED |