mirror of
https://github.com/s3fs-fuse/s3fs-fuse.git
synced 2026-04-25 13:26:00 +03:00
[GH-ISSUE #2496] 1.95 release #1219
Labels
No labels
bug
bug
dataloss
duplicate
enhancement
feature request
help wanted
invalid
need info
performance
pull-request
question
question
testing
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
starred/s3fs-fuse#1219
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
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?
Originally created by @gaul on GitHub (Jul 12, 2024).
Original GitHub issue: https://github.com/s3fs-fuse/s3fs-fuse/issues/2496
Let's run one more release that is compatible with CentOS 7 since we will drop support for it in #2469. I would like to finish off work on the static lock checking since this seems to reveal a bunch of concurrency bugs but otherwise I don't think we have any other outstanding PRs.
@ggtakec commented on GitHub (Jul 12, 2024):
I agree to the 1.95 release.
Let's make this the last version for CentOS 7.
@juliogonzalez
Is it possible to release for CentOS 7?
My test code that I created for CentOS 7 is no longer buildable on Github Actions, so I can't test it with CI.
@juliogonzalez commented on GitHub (Jul 12, 2024):
Possible: Yes
Desiderable: Not so much without tests, IMHO.
I wil try to have a look in the next days to see if I can find why this is failing, as it should not. My suspect is something around
uses: actions/checkout@v3, but not sure right now of what's going on. Maybe it won't be hard to figure it out.@ggtakec commented on GitHub (Jul 12, 2024):
@juliogonzalez
Sorry for my poor explanation.
The reason it's failing is because Github Actions' actions/checkout@v3 can't checkout the source code, and this is a problem specific to GHA.
(It worked right after EOL, but started failing after that)
There should be no problem building the source code.
@juliogonzalez commented on GitHub (Jul 12, 2024):
I see.
But if you create 1.95 from master (and not from the point where you branched to try fix the tests for CentOS7), that would mean releasing with commits that were not really tested for CentOS7. And in this case it would be better to avoid the release.
Let me play a little bit with your branch to see if I can get a fix, before we take a decision. I will come back to you in a few days, probably during the weekend, or shortly after.
@ggtakec commented on GitHub (Jul 12, 2024):
I posted PR #2497 to restore CentOS 7 to CI(Github Actions).
@juliogonzalez
The reason why
actions/checkoutexecution was failing was because it was changed to usenodejs 20(CentOS 7 needsnodejs 16).To avoid this, the
ACTIONS_ALLOW_USE_UNSECURE_NODE_VERSION(=true)environment variable is set.Reference: https://github.com/actions/checkout/issues/1809
@gaul
Could you review this PR #2497 ?
The code added in this PR (only CI-related) can be deleted after the release.
I would like to merge this PR and prepare it for release 1.95.
@gaul commented on GitHub (Oct 16, 2024):
We have a few outstanding locking fixes and the shared_ptr PR but does anything else block 1.95?
@juliogonzalez commented on GitHub (Oct 16, 2024):
Is the plan about this being the last CentOS7 release still valid?
If that's the case, I am not sure EPEL7 is going to accept it. But in any case I will make sure the package still builds for CentOS7 and clones so anyone can easily build 1.95.
@gaul commented on GitHub (Oct 16, 2024):
We haven't merged anything that will break CentOS 7 so 1.95 should be compatible. But I would like to explore FUSE3 and C++17 for 1.96 which will break CentOS 7.
@gaul commented on GitHub (Oct 25, 2024):
Does anything block the release? I suggest these release notes:
ipresolveoption to select IPv4- or IPv6-only@ggtakec commented on GitHub (Oct 25, 2024):
I would like to fix the part I commented on in https://github.com/s3fs-fuse/s3fs-fuse/pull/2569#discussion_r1815840691 .
If I need to make the fix quickly and separate it from #2569, I will immediately submit a PR to fix just that part.
@ggtakec commented on GitHub (Oct 26, 2024):
Congratulations on the v1.95 release
@juliogonzalez commented on GitHub (Oct 27, 2024):
Tracking for RPM submissions
Git main repository
openSUSE
As usual, one accepted this will be submitted to openSUSE Tumbleweed and derivates (such as openSUSE MicroOS), and to new openSUSE Leap and SUSE Linux Enterprise versions.
Existing openSUSE Leap and SUSE Linux Enterprise version will not get unless someone makes a formal request.
Fedora/EPEL
CentOS7 system (or any clone)
Or, as an alternative, anyone can download it from my CI: https://jenkins.juliogonzalez.es/job/s3fs-fuse-rpm-build/DISTRO=centos7,label=docker/56/
@gaul commented on GitHub (Oct 27, 2024):
Thanks as always @juliogonzalez! I will try the Fedora 41 package when it reaches testing.
@juliogonzalez commented on GitHub (Oct 27, 2024):
Done, see above.
Regarding CentOS7 and clones: The submission for EPEL7 could not be prepared (details above), but still anyone can freely and easily build for CentOS7 and clones using my repository.
@juliogonzalez commented on GitHub (Nov 6, 2024):
1.95 now available at:
I still could not push 1.95 (and neither 1.94) to https://build.opensuse.org/project/show/home:juliogonzalez:s3fs-fuse (for Debian/Ubuntu, as non-official package) as https://packages.debian.org/search?keywords=s3fs is still at 1.93 for Debian sid.
@juliogonzalez commented on GitHub (Nov 6, 2024):
@gaul I see you contributed to the Debian package https://salsa.debian.org/debian/s3fs-fuse/-/commits/debian/?ref_type=HEADS, so apparently one doesn't need to be maintainer to send MRs to that repo. I will see what I can do this weekend (sadly I don't have enough time to become a Debian maintainer for s3fs as well, but maybe I can easy things for Mattia Rizzolo)
@gaul commented on GitHub (Nov 6, 2024):
I don't exactly understand how Debian manages releases but this seems to have the entire git history which is why Takeshi and my names show up. I file issues with them sometimes to upgrade to reduce stale bugs filed but haven't done so recently.
@juliogonzalez commented on GitHub (Nov 6, 2024):
You're right. Seems Debian gets the whole git history. I guess they're rebasing changes from https://github.com/s3fs-fuse/s3fs-fuse then adding what Debian needs on top.
As of this moment seems they don't have any bug report, maybe you can send them the request to upgrade to 1.95 on Debian sid?
That way it will at least go to the next Debian and Ubuntu versions, and as soon as they have it in Debian sid, I can provide the (unofficial) packages via https://build.opensuse.org/project/show/home:juliogonzalez:s3fs-fuse
@gaul commented on GitHub (Nov 6, 2024):
I opened this issue to upgrade Debian: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1086865
@juliogonzalez commented on GitHub (Apr 15, 2025):
I bring some news to this issue about CentOS7 and family.
As explained at https://github.com/s3fs-fuse/s3fs-fuse/issues/2496#issuecomment-2440117118, I could not provide packages for EPEL7 back then, as it was already End of Life.
However, since I got contacted by someone to see if I could provide EPEL7 packages (which is, again, not possible), I decided to build it at the same place where I was already building unofficial packages for Debian and Ubuntu:
https://build.opensuse.org/project/show/home:juliogonzalez:s3fs-fuse
And downloads available at:
https://download.opensuse.org/repositories/home:/juliogonzalez:/s3fs-fuse/EL7/
Thing to keep in mind
x86_64,ppc64leandaarch64was failing. So only those architectures are available.@gaul commented on GitHub (Apr 15, 2025):
What kind of build failures? Occasionally 32-bit compatibility issues creep in but those are easy to fix. Raspberry Pi/armv7 is the only real reason to fix these.
@juliogonzalez commented on GitHub (Apr 15, 2025):
Strictly speaking I just noticed inspecting https://build.opensuse.org/projects/Fedora:EPEL:7/meta that is not there
arm7vlFor
arm7vl, I was getting:For
s390xthe same:But this is normal, as those architectures were not imported to the Open Build Service.
Seems
s390xwas never supported on CentOS7. And as for ARM 32 bits (àrm7vl`) I see https://blog.centos.org/2016/01/alternative-architectures-abound-in-centos-7-1511/, so technically speaking it could be added to the Open Build Service. But TBH I don't fill like disturbing the Open Build Service for an EoL OS :-)Of course if someone wants to use an old raspberry in 32 bits mode with CentOS7 (new raspberries support 64 bits), they can still build the RPM manually from https://github.com/juliogonzalez/s3fs-fuse-rpm/tree/1.95-1, so I think this solution is acceptable.
@gaul commented on GitHub (Apr 15, 2025):
OK it's not a s3fs source issue then. I don't have the entire Raspberry Pi situation in my head but even my RPi4 running bookworm defaults to a 64-bit kernel but 32-bit userland due to perceived memory savings.
@juliogonzalez commented on GitHub (Apr 15, 2025):
Right, not a s3fs issue. If someone wants to build locally using the SPEC, they can still do it and should work hopefully just fine (I'd try, but all my Raspberries are running Rasperry Pi OS 32 bits for now :-))