fuel-dev team mailing list archive

[Mike Scherbakov <mscherbakov@xxxxxxxxxxxx>]

A few more things were nailed down during these hours.

I've moved https://bugs.launchpad.net/fuel/+bug/1284261 to 5.0, I hope no
objections to it.

OpenVSwitch with bonding sometimes stops accepting
ARPs<https://bugs.launchpad.net/bugs/1272842> -
Critical, but looks like we can't configure LACP on our switches.
Active-backup mode works fine.
OSError: [Errno 13] Permission denied: for murano
dashboard<https://bugs.launchpad.net/bugs/1282613> -
should be already fixed by package update, waiting for someone to confirm
Murano has too open dirs (o+w) <https://bugs.launchpad.net/bugs/1284574> -
security issue. Waiting for patches from Murano team
Glance Error: Could not find filesystem_store_datadir in configuration
options. <https://bugs.launchpad.net/bugs/1284236> - waiting from Matt more
details on this issue, but looks like it low-priority glance caching issue

I believe we can call for code freeze for all components except Murano. It
means that we will still need to merge trivial changes into Murano DEB
specs, and start accepting testing of Murano only after it is done.
For all other pieces of Fuel I believe we can start acceptance testing.

Any objections?


On Tue, Feb 25, 2014 at 3:42 PM, Mike Scherbakov
<mscherbakov@xxxxxxxxxxxx>wrote:

> Folks,
> few more things were tested and merged. Please pay high attention to every
> new bug created now, and try to triage immediately - we need to know
> whether it's low priority and can be deferred to 5.0 or critical, which
> must be addressed in 4.1.
>
> *We are one step away from calling for a code-freeze.*
> Largest thing to verify is kernel 3.10 (
> https://bugs.launchpad.net/fuel/+bug/1263072,
> https://bugs.launchpad.net/fuel/+bug/1282493)
>
> Needs to be triaged/try/comments provided:
> https://bugs.launchpad.net/fuel/+bug/1284092
> https://bugs.launchpad.net/fuel/+bug/1284236
> https://bugs.launchpad.net/fuel/+bug/1284177
>
> Update about bonding is needed from Andrey:
> https://bugs.launchpad.net/fuel/+bug/1272842
>
> Murano-related issues:
> https://bugs.launchpad.net/fuel/+bug/1282613 - Critical
> https://bugs.launchpad.net/fuel/+bug/1284574
>
> Other:
> https://bugs.launchpad.net/fuel/+bug/1284571 - slow getting of master IP.
> Let's discuss if it's Ok to merge in 4.1.
>
> Thanks,
>
>
> On Tue, Feb 25, 2014 at 2:39 PM, Aleksey Kasatkin <akasatkin@xxxxxxxxxxxx>wrote:
>
>> >>> #1283805 In fuelmenu, if you set IP for master, static pool starts
>> >>> from same IP -- merged,
>> This particular problem is fixed, ticket can be closed.
>> Andrey pointed another problem:
>> https://bugs.launchpad.net/fuel/+bug/1283805/comments/11
>> that can be discussed within separate ticket.
>>
>>
>> Aleksey Kasatkin
>>
>> S. Software Developer | Mirantis, Inc. | http://www.mirantis.com
>> cell: +380938330852 | skype: alexeyk_ru
>>
>>
>> On Tue, Feb 25, 2014 at 8:44 AM, Andrew Woodward <awoodward@xxxxxxxxxxxx>wrote:
>>
>>> I agree kernel-lt-firmware and kernel-lt-headers conflict, headers can
>>> only be installed by the package section because it has deps that it won't
>>> resolve otherwise, but kernel-lt wont replace kernel in that section, hence
>>> the hokey adding it removing the ones we don't want and double checking we
>>> have the ones we do.
>>>
>>> Created https://bugs.launchpad.net/fuel/+bug/1284482 to track defect
>>>
>>>
>>> On Mon, Feb 24, 2014 at 9:57 PM, Matthew Mosesohn <
>>> mmosesohn@xxxxxxxxxxxx> wrote:
>>>
>>>> I'll test the updated CentOS kernel for you today and review.
>>>>
>>>> I see why you had to do the ugly hack for kernel vs kernel-ml, and
>>>> normally this is handled via "conflicts" or "obsoletes" flags. I know
>>>> we can't put both kernels in the same repo with obsoletes flag, but
>>>> maybe we can make it blend with conflicts instead.  If this code
>>>> works, let's ship it for 4.1 since we're down to the wire, but I'd
>>>> like to refactor it into something a little less hackish and doesn't
>>>> make systems complain.
>>>>
>>>> On Tue, Feb 25, 2014 at 9:40 AM, Andrew Woodward <
>>>> awoodward@xxxxxxxxxxxx> wrote:
>>>> > https://review.openstack.org/#/c/76090/ for fuel-web to convert
>>>> names from
>>>> > kernel_ml to kernel_lt
>>>> >
>>>> > https://review.openstack.org/#/c/76089/ to fix the package names and
>>>> crazy
>>>> > amount of work that is required to effectively get anaconda to change
>>>> the
>>>> > kernel before it finishes.
>>>> >
>>>> > https://review.openstack.org/76092 to fix the rpm requires and
>>>> should be
>>>> > merged after OSCI has all packages.
>>>> >
>>>> > Andrew.
>>>> >
>>>> >
>>>> >
>>>> > On Mon, Feb 24, 2014 at 8:41 PM, Dmitry Borodaenko
>>>> > <dborodaenko@xxxxxxxxxxxx> wrote:
>>>> >>
>>>> >> Team,
>>>> >>
>>>> >> We are almost ready to announce code freeze for 4.1. Unfortunately,
>>>> we
>>>> >> couldn't merge everything that we need today, so we can't freeze just
>>>> >> yet. The good news is that we didn't find any major regressions as of
>>>> >> ISO #195, and there is only a handful of problems that we still need
>>>> >> to address:
>>>> >>
>>>> >> 1) Bonding
>>>> >> 2) Kernel 3.10 for CentOS
>>>> >> 3) Post-deploy orchestration workaround
>>>> >>
>>>> >> I believe that other outstanding issues are not critical enough to be
>>>> >> considered release blockers. If you know of any other bugs that must
>>>> >> be fixed in 4.1, please speak up now. Otherwise, I propose a soft
>>>> code
>>>> >> freeze: lets not merge any other changes except fixes for the above
>>>> >> three issues, and declare code freeze as soon as these are fixed.
>>>> >>
>>>> >> Bonding:
>>>> >>
>>>> >> There was a lot of confusion today about #1272842 (OpenVSwitch with
>>>> >> bonding sometimes stops accepting ARPs). We thought we have a way to
>>>> >> reliably reproduce this bug, but all we found was a lot of problems
>>>> >> with our test methodology for NIC bonding, and a real and easily
>>>> fixed
>>>> >> bug #1284384 (Remove balance-tcp bond mode option). Our conclusion so
>>>> >> far is that bonding isn't going to work in our devops/kvm based test
>>>> >> environment, and we will need a real switch with LACP to have another
>>>> >> go at trying to reproduce the bonding/ARP problem. If that is the
>>>> case
>>>> >> and we can't really reproduce #1272842 tomorrow, I propose that we
>>>> fix
>>>> >> #1284384 and leave #1272842 as a known limitation.
>>>> >>
>>>> >> Kernel 3.10 for CentOS
>>>> >>
>>>> >> Andrew's fuel-library patch was merged prematurely (before the
>>>> >> required kernel rpm's where created and added to the package
>>>> >> repositories). Andrew is close to getting his Kickstart script
>>>> changes
>>>> >> work with the new kernel-lt packages, but unfortunately the new
>>>> >> packages still miss the matching firmware binary package, he had to
>>>> >> download that package from the upstream site to manually test his
>>>> >> changes. I think we need some close coordination between Andrew,
>>>> Matt,
>>>> >> and OSCI team tomorrow morning Pacific time to tie all loose ends
>>>> >> here: we need the firmware package from OSCI, and it would be very
>>>> >> nice if Matt could review Andrew's fixes before we merge an updated
>>>> >> version of them.
>>>> >>
>>>> >> Post-deploy orchestration workaround
>>>> >>
>>>> >> There's a -1 from QA on https://review.openstack.org/#/c/75389/ , it
>>>> >> can't be merged until that is resolved. There's hope that this was a
>>>> >> one-off test failure unrelated to the code change: Aleksandr Didenko
>>>> >> commented that at least the /etc/hosts part worked fine, although he
>>>> >> wasn't able to check the radosgw side of the fix. If that's the case,
>>>> >> we should be able to merge this tomorrow.
>>>> >>
>>>> >> Other critical and high priority bugs:
>>>> >>
>>>> >> #1284002 Request for HA deployment -- the consensus this morning was
>>>> >> this this bug report is invalid, I think we need a document on how to
>>>> >> manually test Neutron, something similar to
>>>> >> fuel-library/deployment/puppet/ceph/README.md, so that everyone can
>>>> >> agree and what is and isn't supposed to work. having an OSTF test
>>>> >> script helps, but is not enough, there should be a document explains
>>>> >> the test steps, and may include checks that are not easily scripted
>>>> >>
>>>> >> #1263934 fuelmenu NTP settings prevent me from proceeding -- merged
>>>> >>
>>>> >> #1267431 Get rid of hardcoded "admin" tenant in puppet modules -- a
>>>> >> short-term fix was merged, the rest should be postponed to 5.0
>>>> >>
>>>> >> #1283805 In fuelmenu, if you set IP for master, static pool starts
>>>> >> from same IP -- merged, not sure if there's more work left for 5.0 or
>>>> >> this can be closed?
>>>> >>
>>>> >> #1283812 Stop deploy tasks fails if any node was inaccessible --
>>>> >> doesn't look like this can be done in 4.1, unless a really safe and
>>>> >> non-intrusive fix is proposed tomorrow
>>>> >>
>>>> >> #1284092 Instances can't get metadata and IP addresses -- didn't have
>>>> >> time to triage, is this confirmed?
>>>> >>
>>>> >> #1284177 OS not erased from node after cluster is deleted -- a
>>>> >> duplicate of #1283825?
>>>> >>
>>>> >> #1284236 Glance Error: Could not find filesystem_store_datadir in
>>>> >> configuration options -- didn't have time to triage
>>>> >>
>>>> >> #1284261 Astute must compare a sets of UIDs when collect data from
>>>> >> netprobe.py -- proposed fix is a simple one-liner, can this be done
>>>> >> tomorrow?
>>>> >>
>>>> >> #1284270 Please, add notification for user of fuel-cli about custom
>>>> >> settings -- I think this should be postpoed to 5.0
>>>> >>
>>>> >> #1284193 ImageNotFound: error opening image
>>>> >> /var/lib/nova/instances/_base/ -- Nova bug (live migrations work but
>>>> >> non-live migrations don't work with ephemeral RBD backend), postponed
>>>> >> to 5.0
>>>> >>
>>>> >> --
>>>> >> Dmitry Borodaenko
>>>> >>
>>>> >> --
>>>> >> You received this message because you are subscribed to the Google
>>>> Groups
>>>> >> "fuel-core-team" group.
>>>> >> To unsubscribe from this group and stop receiving emails from it,
>>>> send an
>>>> >> email to fuel-core-team+unsubscribe@xxxxxxxxxxxx.
>>>> >> For more options, visit
>>>> >> https://groups.google.com/a/mirantis.com/groups/opt_out.
>>>> >
>>>> >
>>>> > --
>>>> > You received this message because you are subscribed to the Google
>>>> Groups
>>>> > "fuel-core-team" group.
>>>> > To unsubscribe from this group and stop receiving emails from it,
>>>> send an
>>>> > email to fuel-core-team+unsubscribe@xxxxxxxxxxxx.
>>>> > For more options, visit
>>>> > https://groups.google.com/a/mirantis.com/groups/opt_out.
>>>>
>>>
>>>  --
>>> You received this message because you are subscribed to the Google
>>> Groups "fuel-core-team" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an email to fuel-core-team+unsubscribe@xxxxxxxxxxxx.
>>> For more options, visit
>>> https://groups.google.com/a/mirantis.com/groups/opt_out.
>>>
>>
>>  --
>> You received this message because you are subscribed to the Google Groups
>> "fuel-core-team" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to fuel-core-team+unsubscribe@xxxxxxxxxxxx.
>> For more options, visit
>> https://groups.google.com/a/mirantis.com/groups/opt_out.
>>
>
>
>
> --
> Mike Scherbakov
> #mihgen
>



-- 
Mike Scherbakov
#mihgen