ceph-object-corpus test coverage

List overview All Threads
Download

newer

older

el9 build nautilus

reef 18.2.3 QE validation status

Casey Bodley

16 Feb 2024 16 Feb '24

9:12 p.m.

this ceph-object-corpus repo is the basis of our ceph-dencoder test src/test/encoding/readable.sh, which verifies that we can still decode all of the data structures encoded by older ceph versions i'd like to raise awareness that this ceph-object-corpus repo hasn't been updated with new encodings since pacific 16.2.0, so we're missing important regression test coverage since then Nitzan prepared the encodings for reef 18.2.0 in https://github.com/ceph/ceph-object-corpus/pull/17, but those haven't merged yet. i had opened https://github.com/ceph/ceph/pull/54735 to test that, but 'make check' identified failures like:

...

The following tests FAILED: 147 - readable.sh (Failed) **** reencode of /home/jenkins-build/build/workspace/ceph-pull-requests/ceph-object-corpus/archive/18.2.0/objects/chunk_refs_t/ccb69d9ecd572c1f6ed9598899773cf1 resulted in a different dump ****

can we find a way to prioritize this? it would be great to have these reef encodings while we're validating the squid release

Show replies by date

Gregory Farnum

17 Feb 17 Feb

10:01 a.m.

I think we should make updating this a blocker for the squid release. Maintaining ceph-object-corpus has always been very ad-hoc so we should perhaps add it to release checklists. If there are specific failures in the object corpus tests, those should be directed to the relevant component leads/recent updaters, right? -Greg On Fri, Feb 16, 2024 at 10:12 AM Casey Bodley <cbodley(a)redhat.com> wrote:

...

The following tests FAILED: 147 - readable.sh (Failed) **** reencode of

/home/jenkins-build/build/workspace/ceph-pull-requests/ceph-object-corpus/archive/18.2.0/objects/chunk_refs_t/ccb69d9ecd572c1f6ed9598899773cf1 resulted in a different dump **** can we find a way to prioritize this? it would be great to have these reef encodings while we're validating the squid release _______________________________________________ Dev mailing list -- dev(a)ceph.io To unsubscribe send an email to dev-leave(a)ceph.io

Nitzan Mordechai

18 Feb 18 Feb

8:35 a.m.

actually. the checklist do have corpus tests listed: generate new object corpus for encoding/decoding tests i'll ask Neha or someone else to approve the PR that is supposed to fix it ________________________________ From: Gregory Farnum <gfarnum(a)redhat.com> Sent: Saturday, February 17, 2024 9:01 AM To: Casey Bodley <cbodley(a)redhat.com> Cc: Nitzan Mordechai <nmordech(a)redhat.com>om>; dev(a)ceph.io <dev(a)ceph.io> Subject: [EXTERNAL] Re: ceph-object-corpus test coverage I think we should make updating this a blocker for the squid release. Maintaining ceph-object-corpus has always been very ad-hoc so we should perhaps add it to release checklists. If there are specific failures in the object corpus tests, those ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. <https://us-phishalarm-ewt.proofpoint.com/EWT/v1/PjiDSg!12-vrJE_wRb0UGibti1tLhwGRQx9mkw0ZCvVIFehRyDWuLnrR44lCN4hBflALgWUO_EP_eul_pKmWlkN5SO1VvDm_G1ldMWblLIEgfCUxc_-vhiQop4fxqWE$> Report Suspicious ZjQcmQRYFpfptBannerEnd I think we should make updating this a blocker for the squid release. Maintaining ceph-object-corpus has always been very ad-hoc so we should perhaps add it to release checklists. If there are specific failures in the object corpus tests, those should be directed to the relevant component leads/recent updaters, right? -Greg On Fri, Feb 16, 2024 at 10:12 AM Casey Bodley <cbodley@redhat.com<mailto:cbodley@redhat.com>> wrote: this ceph-object-corpus repo is the basis of our ceph-dencoder test src/test/encoding/readable.sh, which verifies that we can still decode all of the data structures encoded by older ceph versions i'd like to raise awareness that this ceph-object-corpus repo hasn't been updated with new encodings since pacific 16.2.0, so we're missing important regression test coverage since then Nitzan prepared the encodings for reef 18.2.0 in https://github.com/ceph/ceph-object-corpus/pull/17<https://github.com/ce…17>, but those haven't merged yet. i had opened https://github.com/ceph/ceph/pull/54735<https://github.com/ceph/ceph/pul… to test that, but 'make check' identified failures like:

...

can we find a way to prioritize this? it would be great to have these reef encodings while we're validating the squid release _______________________________________________ Dev mailing list -- dev@ceph.io<mailto:dev@ceph.io> To unsubscribe send an email to dev-leave@ceph.io<mailto:dev-leave@ceph.io>

Casey Bodley

3 Apr 3 Apr

8:54 p.m.

the updated ceph-object-corpus submodule is finally passing tests in https://github.com/ceph/ceph/pull/54735 but needs an approval from the core team On Sun, Feb 18, 2024 at 12:35 AM Nitzan Mordechai <nmordech(a)ibm.com> wrote:

...

actually. the checklist do have corpus tests listed: generate new object corpus for encoding/decoding tests i'll ask Neha or someone else to approve the PR that is supposed to fix it ________________________________ From: Gregory Farnum <gfarnum(a)redhat.com> Sent: Saturday, February 17, 2024 9:01 AM To: Casey Bodley <cbodley(a)redhat.com> Cc: Nitzan Mordechai <nmordech(a)redhat.com>om>; dev(a)ceph.io <dev(a)ceph.io> Subject: [EXTERNAL] Re: ceph-object-corpus test coverage I think we should make updating this a blocker for the squid release. Maintaining ceph-object-corpus has always been very ad-hoc so we should perhaps add it to release checklists. If there are specific failures in the object corpus tests, those ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. Report Suspicious ZjQcmQRYFpfptBannerEnd I think we should make updating this a blocker for the squid release. Maintaining ceph-object-corpus has always been very ad-hoc so we should perhaps add it to release checklists. If there are specific failures in the object corpus tests, those should be directed to the relevant component leads/recent updaters, right? -Greg On Fri, Feb 16, 2024 at 10:12 AM Casey Bodley <cbodley(a)redhat.com> wrote: this ceph-object-corpus repo is the basis of our ceph-dencoder test src/test/encoding/readable.sh, which verifies that we can still decode all of the data structures encoded by older ceph versions i'd like to raise awareness that this ceph-object-corpus repo hasn't been updated with new encodings since pacific 16.2.0, so we're missing important regression test coverage since then Nitzan prepared the encodings for reef 18.2.0 in https://github.com/ceph/ceph-object-corpus/pull/17, but those haven't merged yet. i had opened https://github.com/ceph/ceph/pull/54735 to test that, but 'make check' identified failures like:

can we find a way to prioritize this? it would be great to have these reef encodings while we're validating the squid release _______________________________________________ Dev mailing list -- dev(a)ceph.io To unsubscribe send an email to dev-leave(a)ceph.io

Radoslaw Zarzynski

9:18 p.m.

Done! On Wed, Apr 3, 2024 at 7:55 PM Casey Bodley <cbodley(a)redhat.com> wrote:

...

actually. the checklist do have corpus tests listed: generate new object corpus for encoding/decoding tests i'll ask Neha or someone else to approve the PR that is supposed to fix it ________________________________ From: Gregory Farnum <gfarnum(a)redhat.com> Sent: Saturday, February 17, 2024 9:01 AM To: Casey Bodley <cbodley(a)redhat.com> Cc: Nitzan Mordechai <nmordech(a)redhat.com>om>; dev(a)ceph.io <dev(a)ceph.io> Subject: [EXTERNAL] Re: ceph-object-corpus test coverage I think we should make updating this a blocker for the squid release. Maintaining ceph-object-corpus has always been very ad-hoc so we should perhaps add it to release checklists. If there are specific failures in the object corpus tests, those ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. Report Suspicious ZjQcmQRYFpfptBannerEnd I think we should make updating this a blocker for the squid release. Maintaining ceph-object-corpus has always been very ad-hoc so we should perhaps add it to release checklists. If there are specific failures in the object corpus tests, those should be directed to the relevant component leads/recent updaters, right? -Greg On Fri, Feb 16, 2024 at 10:12 AM Casey Bodley <cbodley(a)redhat.com> wrote: this ceph-object-corpus repo is the basis of our ceph-dencoder test src/test/encoding/readable.sh, which verifies that we can still decode all of the data structures encoded by older ceph versions i'd like to raise awareness that this ceph-object-corpus repo hasn't been updated with new encodings since pacific 16.2.0, so we're missing important regression test coverage since then Nitzan prepared the encodings for reef 18.2.0 in https://github.com/ceph/ceph-object-corpus/pull/17, but those haven't merged yet. i had opened https://github.com/ceph/ceph/pull/54735 to test that, but 'make check' identified failures like:

can we find a way to prioritize this? it would be great to have these reef encodings while we're validating the squid release _______________________________________________ Dev mailing list -- dev(a)ceph.io To unsubscribe send an email to dev-leave(a)ceph.io

_______________________________________________ Dev mailing list -- dev(a)ceph.io To unsubscribe send an email to dev-leave(a)ceph.io

Casey Bodley

18 Apr 18 Apr

10:05 p.m.

the squid backport merged in https://github.com/ceph/ceph/pull/56676, so we're all done here. thanks Nitzan! we'll want to add similar coverage for 19.2.0 to main once squid ships On Wed, Apr 3, 2024 at 2:18 PM Radoslaw Zarzynski <rzarzyns(a)redhat.com> wrote:

...

Done! On Wed, Apr 3, 2024 at 7:55 PM Casey Bodley <cbodley(a)redhat.com> wrote:

actually. the checklist do have corpus tests listed: generate new object corpus for encoding/decoding tests i'll ask Neha or someone else to approve the PR that is supposed to fix it ________________________________ From: Gregory Farnum <gfarnum(a)redhat.com> Sent: Saturday, February 17, 2024 9:01 AM To: Casey Bodley <cbodley(a)redhat.com> Cc: Nitzan Mordechai <nmordech(a)redhat.com>om>; dev(a)ceph.io <dev(a)ceph.io> Subject: [EXTERNAL] Re: ceph-object-corpus test coverage I think we should make updating this a blocker for the squid release. Maintaining ceph-object-corpus has always been very ad-hoc so we should perhaps add it to release checklists. If there are specific failures in the object corpus tests, those ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. Report Suspicious ZjQcmQRYFpfptBannerEnd I think we should make updating this a blocker for the squid release. Maintaining ceph-object-corpus has always been very ad-hoc so we should perhaps add it to release checklists. If there are specific failures in the object corpus tests, those should be directed to the relevant component leads/recent updaters, right? -Greg On Fri, Feb 16, 2024 at 10:12 AM Casey Bodley <cbodley(a)redhat.com> wrote: this ceph-object-corpus repo is the basis of our ceph-dencoder test src/test/encoding/readable.sh, which verifies that we can still decode all of the data structures encoded by older ceph versions i'd like to raise awareness that this ceph-object-corpus repo hasn't been updated with new encodings since pacific 16.2.0, so we're missing important regression test coverage since then Nitzan prepared the encodings for reef 18.2.0 in https://github.com/ceph/ceph-object-corpus/pull/17, but those haven't merged yet. i had opened https://github.com/ceph/ceph/pull/54735 to test that, but 'make check' identified failures like:

can we find a way to prioritize this? it would be great to have these reef encodings while we're validating the squid release _______________________________________________ Dev mailing list -- dev(a)ceph.io To unsubscribe send an email to dev-leave(a)ceph.io

_______________________________________________ Dev mailing list -- dev(a)ceph.io To unsubscribe send an email to dev-leave(a)ceph.io

days inactive

days old

dev@ceph.io

Manage subscription

5 comments

4 participants

tags (0)

participants (4)

Casey Bodley
Gregory Farnum
Nitzan Mordechai
Radoslaw Zarzynski