quality/tickets

Fork 0

Re-evaluate the list of release-blocking storage interfaces #890

New issue

Open

opened 2026-04-14 16:18:41 +00:00 by kparal · 10 comments

kparal commented

2026-04-14 16:18:41 +00:00

Owner

Last year we decided that we want to:

Discuss whether we still need to block on so many storage interfaces

But that never got done. And now that our team has lost capacity to test advanced Server hardware (SAS, HWRAID, iSCSI and FCoE; optionally Multipath), this topic is more important than before.

So go through https://fedoraproject.org/wiki/Template:Installation_test_matrix#Storage_devices and discuss with the Server WG and other involved stakeholders, whether all of that is still release-critical (i.e. we should still block on it), or whether we can reduce the list. According to the output of that discussion, submit proposals, update release criteria and wiki matrices.

Progress checklist:

Discuss with Server team
Create a list of proposed changes and send them out
Participate in the discussion, finalize the changes
Implement the changes

Last year we [decided](https://discussion.fedoraproject.org/t/fedora-quality-scope-reduction-announcement-and-summary/160523) that we want to: > Discuss whether we still need to block on so many storage interfaces But that never got done. And now that our team has lost capacity to test advanced Server hardware (SAS, HWRAID, iSCSI and FCoE; optionally Multipath), this topic is more important than before. So go through https://fedoraproject.org/wiki/Template:Installation_test_matrix#Storage_devices and discuss with the Server WG and other involved stakeholders, whether all of that is still release-critical (i.e. we should still block on it), or whether we can reduce the list. According to the output of that discussion, submit proposals, update release criteria and wiki matrices. --- **Progress checklist:** - [x] Discuss with Server team - [ ] Create a list of proposed changes and send them out - [ ] Participate in the discussion, finalize the changes - [ ] Implement the changes

kparal added the

labels

2026-04-14 16:18:41 +00:00

kparal referenced this issue

2026-04-30 17:09:06 +00:00

Plan what to do about advanced storage testing #889

kparal added this to the Fedora 45 milestone

2026-04-30 17:09:16 +00:00

psklenar was assigned by kparal

2026-05-05 08:32:03 +00:00

kparal added this to the Sprint 9 project

2026-05-05 08:32:06 +00:00

kparal commented

2026-05-05 08:37:36 +00:00

Author

Owner

These release criteria are likely related:
https://fedoraproject.org/wiki/Basic_Release_Criteria#Storage_interfaces
https://fedoraproject.org/wiki/Fedora_45_Beta_Release_Criteria#Hardware_RAID
https://fedoraproject.org/wiki/Fedora_45_Final_Release_Criteria#Firmware_RAID
https://fedoraproject.org/wiki/Fedora_45_Final_Release_Criteria#network-attached-storage

These release criteria are likely related: https://fedoraproject.org/wiki/Basic_Release_Criteria#Storage_interfaces https://fedoraproject.org/wiki/Fedora_45_Beta_Release_Criteria#Hardware_RAID https://fedoraproject.org/wiki/Fedora_45_Final_Release_Criteria#Firmware_RAID https://fedoraproject.org/wiki/Fedora_45_Final_Release_Criteria#network-attached-storage

kparal commented

2026-05-05 08:42:00 +00:00

Author

Owner

Probably create a Server ticket or ask in the server mailing list.

Probably create a [Server ticket](https://forge.fedoraproject.org/server) or ask in the server mailing list.

psklenar referenced this issue

2026-05-05 10:05:58 +00:00

Plan what to do about advanced storage testing #889

psklenar commented

2026-05-07 15:35:52 +00:00

Owner

there is summary after looking into Bugzilla and today technology:

MUST TEST (Keep Release-Blocking)

NVMe
SATA - Still dominant for HDDs/budget SSDs
VirtIO
Software RAID - More common than hardware RAID,

REQUIRES SERVER WG CONSULTATION
(will check with team)

iSCSI- Enterprise SAN deployments, currently Final blocking - can be tested by two VM
Multipath - Advanced HA configurations, currently optional, can be tested by two disks with the same serial number

REDUCE TO BEST-EFFORT

SAS - Limited hw but can be found in beaker
FCoE** - Limited hw, didnt find around
Hardware RAID- Limited hw , can be found in beaker

there is summary after looking into Bugzilla and today technology: **MUST TEST (Keep Release-Blocking)** NVMe SATA - Still dominant for HDDs/budget SSDs VirtIO Software RAID - More common than hardware RAID, **REQUIRES SERVER WG CONSULTATION** _(will check with team)_ iSCSI- Enterprise SAN deployments, currently Final blocking - can be tested by two VM Multipath - Advanced HA configurations, currently optional, can be tested by two disks with the same serial number **REDUCE TO BEST-EFFORT** SAS - Limited hw but can be found in beaker FCoE** - Limited hw, didnt find around Hardware RAID- Limited hw , can be found in beaker

kparal commented

2026-05-07 16:10:36 +00:00

Author

Owner

Summary from who, Claude? 🙂️

I'd like have the Server team opinion on all of SAS, HWRAID, iSCSI, FCoE and Multipath. Multipath is optional, so we don't really care that much (we don't have to test it), but the rest of it is currently mandatory. It is up to us to propose a criteria change. The two justification types are: a) we can't test it, or b) we believe it's no longer worth blocking on. So we have to figure out which storage type fits where (or whether we keep it release blocking).

Summary from who, Claude? 🙂️ I'd like have the Server team opinion on all of SAS, HWRAID, iSCSI, FCoE and Multipath. Multipath is optional, so we don't really care that much (we don't have to test it), but the rest of it is currently mandatory. It is up to us to propose a criteria change. The two justification types are: a) we can't test it, or b) we believe it's no longer worth blocking on. So we have to figure out which storage type fits where (or whether we keep it release blocking).

adamwill commented

2026-05-07 18:39:47 +00:00

Owner

Note we already test iSCSI virtualized in openQA, so it's a "solved problem". I have a ticket for doing multipath this way but have never got around to it; in theory that would be a one-time lift and then minimal maintenance / failure review.

I think I looked and couldn't find any practical way to virtualize FCoE testing, but maybe someone else knows something?

Note we already test iSCSI virtualized in openQA, so it's a "solved problem". I have a [ticket](https://forge.fedoraproject.org/quality/os-autoinst-distri-fedora/issues/430) for doing multipath this way but have never got around to it; in theory that would be a one-time lift and then minimal maintenance / failure review. I think I looked and couldn't find any practical way to virtualize FCoE testing, but maybe someone else knows something?

adamwill commented

2026-05-07 18:41:29 +00:00

Owner

Hmm, a bit more googling shows up https://github.com/ipxe/fcoed ...hasn't been touched for six years, but maybe if it works it works?

psklenar commented

2026-05-11 07:59:48 +00:00

Owner

@kparal
yes - I have been using everything I was told to use: Claude and Gemini.
I will ask the "Fedora Server WG" during their next meeting; the nearest one is on Wednesday, May 20th.

@adamwill
iSCSI - good to know we already have it automated;

multipath:

I found that there was a test by RTT/anaconda qa team:
storage-multipath-autopart.ks.in
which is (and will be) stopped due to BZ#1853668
There is also a guide on how to use multipath in a VM by providing fake ONE serial number to TWO disks:
https://redhat.atlassian.net/wiki/spaces/RTT/pages/307364353/Multipath+in+VM
which I tried and resulted in this bz: https://bugzilla.redhat.com/show_bug.cgi?id=2469006#c7
Workstation is buggy, Server is not. Workstation is missing packages.

@kparal yes - I have been using everything I was told to use: Claude and Gemini. I will ask the "Fedora Server WG" during their next meeting; the nearest one is on Wednesday, May 20th. @adamwill iSCSI - good to know we already have it automated; **multipath**: 1. I found that there was a test by RTT/anaconda qa team: [storage-multipath-autopart.ks.in](https://github.com/rhinstaller/kickstart-tests/blob/main/storage-multipath-autopart.ks.in) which is (and will be) stopped due to [BZ#1853668](https://bugzilla.redhat.com/show_bug.cgi?id=1853668) 2. There is also a guide on how to use multipath in a VM by providing fake ONE serial number to TWO disks: https://redhat.atlassian.net/wiki/spaces/RTT/pages/307364353/Multipath+in+VM which I tried and resulted in this bz: https://bugzilla.redhat.com/show_bug.cgi?id=2469006#c7 _Workstation is buggy, Server is not. Workstation is missing packages._

kparal commented

2026-05-11 09:19:57 +00:00

Author

Owner

@adamwill wrote in #890 (comment):

Note we already test iSCSI virtualized in openQA, so it's a "solved problem".

That's great, but the question of "Should we still block on it?" is still valid, isn't it?

@psklenar wrote in #890 (comment):

I will ask the "Fedora Server WG" during their next meeting; the nearest one is on Wednesday, May 20th.

I would prefer us creating a ticket in their tracker way before this date, so they are informed and think about it ahead (and then also discuss it during the meeting). If we don't give them any heads up, their real-time response might not be as useful as we'd like.

@adamwill wrote in https://forge.fedoraproject.org/quality/tickets/issues/890#issuecomment-698705: > Note we already test iSCSI virtualized in openQA, so it's a "solved problem". That's great, but the question of "Should we still block on it?" is still valid, isn't it? @psklenar wrote in https://forge.fedoraproject.org/quality/tickets/issues/890#issuecomment-705749: > I will ask the "Fedora Server WG" during their next meeting; the nearest one is on Wednesday, May 20th. I would prefer us creating a ticket in their tracker way before this date, so they are informed and think about it ahead (and then also discuss it during the meeting). If we don't give them any heads up, their real-time response might not be as useful as we'd like.

psklenar referenced this issue from server/tickets

2026-05-11 09:53:18 +00:00

Re-evaluate the list of release-blocking storage interfaces #196

psklenar commented

2026-05-18 13:41:11 +00:00

Owner

there is some progress:
keep blocking NVMe SATA, VirtIO, iSCSI
remove blocking: PATA SCSI, firmware RAID
+ server WG will waive the blocking status or FCOe and Multipath as long as there is no test equipment available

HW raid needs more time for discussion
SAS not explicitly decided => keeps blocking

there is some progress: keep blocking NVMe SATA, VirtIO, iSCSI remove blocking: PATA SCSI, firmware RAID _+ server WG will waive the blocking status or FCOe and Multipath as long as there is no test equipment available_ HW raid needs more time for discussion SAS not explicitly decided => keeps blocking