sig_core
/
meta
8
0
Issues 13 Pull Requests Projects 5 Releases Wiki Activity

Task: Fix broken packages for Rocky Linux 10 #16

New Issue
Open
opened 2024-03-12 15:56:12 +00:00 by label · 9 comments
label commented 2024-03-12 15:56:12 +00:00
Owner
Copy Link

Below are a list of packages that cannot be built for Rocky Linux 10, and their respective issues.

  • libecpg - Fails randomly on different arches, primarily aarch64. Example build
  • mingw-termcap - Fails with memory related messages - not clear if OOM
  • java-*-openjdk-portable - Macro issues.
  • rust - s390x OOM
  • scipy - (consistent crashing, arch doesn't matter)
  • dotnet8.0 - requires a custom bootstrap. low priority.
  • shim-unsigned-aarch64 - compilation error, not clear where. (solved with efi-srpm-macros update)

Below are a list of packages that cannot be imported due to "+" in the name and the source in which they are being imported from are already gitlabified.

  • libsigc++20
  • libsigc++30
  • perl-Text-Tabs+Wrap
  • memtest86+
Below are a list of packages that cannot be built for Rocky Linux 10, and their respective issues. - [x] libecpg - Fails randomly on different arches, primarily aarch64. [Example build](https://peridot.build.resf.org/e7b83c0a-b514-4903-b739-6943bbb307f7/tasks/5cecbeff-411c-4088-b5ba-a3018aae4ff3) - [x] mingw-termcap - Fails with memory related messages - not clear if OOM - [x] java-*-openjdk-portable - Macro issues. - [x] rust - s390x OOM - [x] scipy - (consistent crashing, arch doesn't matter) - [ ] dotnet8.0 - requires a custom bootstrap. low priority. - [x] shim-unsigned-aarch64 - compilation error, not clear where. (solved with efi-srpm-macros update) Below are a list of packages that cannot be imported due to "+" in the name and the source in which they are being imported from are already gitlabified. - [ ] libsigc++20 - [ ] libsigc++30 - [ ] perl-Text-Tabs+Wrap - [ ] memtest86+
label added this to the Rocky Linux 10 project 2024-03-12 18:22:21 +00:00
neil commented 2024-03-18 21:36:40 +00:00
Owner
Copy Link

mingw-termcap - Fails with memory related messages - not clear if OOM

-- missing unistd.h; need to patch source to include.

>mingw-termcap - Fails with memory related messages - not clear if OOM -- missing unistd.h; need to patch source to include.
label commented 2024-03-19 05:35:39 +00:00
Author
Owner
Copy Link

java-*-openjdk-portable - Macro issues.

Resolved using this patch configuration, due to RPM 4.19 changes.

>java-*-openjdk-portable - Macro issues. Resolved using [this patch configuration](https://git.rockylinux.org/staging/patch/java-21-openjdk-portable/-/blob/r10/ROCKY/CFG/java-21-openjdk.cfg), due to RPM 4.19 changes.
neil commented 2024-03-19 14:44:01 +00:00
Owner
Copy Link

java-*-openjdk-portable - Macro issues.

Resolved using this patch configuration, due to RPM 4.19 changes.

is this going to affect upstream? should we even try to PR a fix or open a ticket about it?

> >java-*-openjdk-portable - Macro issues. > > Resolved using [this patch configuration](https://git.rockylinux.org/staging/patch/java-21-openjdk-portable/-/blob/r10/ROCKY/CFG/java-21-openjdk.cfg), due to RPM 4.19 changes. is this going to affect upstream? should we even try to PR a fix or open a ticket about it?
label commented 2024-03-19 15:38:59 +00:00
Author
Owner
Copy Link

I am not sure.

Fedora employs similar spec changes to their portable packages, so they're already covered. I'm not sure how we could convey to CentOS Stream about this. I see issues with what they currently have:

  • They don't have separate git repositories for the *-portable packages, they instead have the portable spec file as a source file
  • The c10s branch of java-*-openjdk is a copy of Fedora, so the above doesn't apply yet
  • The c8s and c9s branches of java-*-openjdk don't appear to have been updated in a long time
  • Even if the point directly above wasn't true, they purposely build the portables on an older release (e.g. EL7 or EL8) and use that to build the actual package that is shipped to users

I just wonder if they would be receptive to the change on their jira, since this seems to only affect RPM 4.19, which is EL10 (and current Fedora). I don't think it would hurt to try.

I am not sure. Fedora employs similar spec changes to their portable packages, so they're already covered. I'm not sure how we could convey to CentOS Stream about this. I see issues with what they currently have: * They don't have separate git repositories for the *-portable packages, they instead have the portable spec file as a source file * The c10s branch of java-*-openjdk is a copy of Fedora, so the above doesn't apply yet * The c8s and c9s branches of java-*-openjdk don't appear to have been updated in a long time * Even if the point directly above wasn't true, they purposely build the portables on an older release (e.g. EL7 or EL8) and use that to build the actual package that is shipped to users I just wonder if they would be receptive to the change on their jira, since this seems to only affect RPM 4.19, which is EL10 (and current Fedora). I don't think it would hurt to try.
neil commented 2024-03-19 15:59:35 +00:00
Owner
Copy Link

I've put out a question explaining the situation in the CentOS Integration SIG chat -- https://chat.fedoraproject.org/#/room/#centos-integration:fedora.im

I've put out a question explaining the situation in the CentOS Integration SIG chat -- https://chat.fedoraproject.org/#/room/#centos-integration:fedora.im
neil commented 2024-03-19 18:32:32 +00:00
Owner
Copy Link

re: java-*-openjdk -- rpm-build in 10 no longer Provides: /usr/lib/rpm/debugedit -- use binary from debugedit which is already BuildRequired.

re: java-*-openjdk -- rpm-build in 10 no longer Provides: /usr/lib/rpm/debugedit -- use binary from `debugedit` which is already BuildRequired.
label commented 2024-03-21 19:14:53 +00:00
Author
Owner
Copy Link

rust - s390x OOM

This required adding some swap to the current nodes and basically getting lucky. Current plan is to get systems with stronger resources to avoid this issue with not only rust, but other packages that use a ridiculous amount of RAM.

>rust - s390x OOM This required adding some swap to the current nodes and basically getting lucky. Current plan is to get systems with stronger resources to avoid this issue with not only rust, but other packages that use a ridiculous amount of RAM.
label commented 2024-03-21 22:49:03 +00:00
Author
Owner
Copy Link

scipy - (consistent crashing, arch doesn't matter)

It decided to pass this time. disable checks may have not been necessary as it was crashing before even with it.

>scipy - (consistent crashing, arch doesn't matter) It decided to pass this time. disable checks may have not been necessary as it was crashing before even with it.
label commented 2024-03-22 01:30:19 +00:00
Author
Owner
Copy Link

libecpg - Fails randomly on different arches, primarily aarch64

Reducing the number of cpu's used allowed the build to succeed on aarch64 and s390x

>libecpg - Fails randomly on different arches, primarily aarch64 Reducing the number of cpu's used allowed the build to succeed on aarch64 and s390x
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
main
Labels
Clear labels   
Discussion

   
FTBFS

   
Meeting

   
OS_Development

   
Research

   
SIG_Proposal

   
SIG_Request
No Label
Discussion
FTBFS
Meeting
OS_Development
Research
SIG_Proposal
SIG_Request
Milestone
Clear milestone
No items
No Milestone
Projects
Clear projects
No project
Rocky Linux 10
Assignees
Clear assignees
No Assignees
2 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: sig_core/meta#16
No description provided.
Delete Branch "%!s(<nil>)"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?