This bug was fixed in the package klibc - 2.0.4-9ubuntu2.2
---
klibc (2.0.4-9ubuntu2.2) bionic; urgency=medium
[ Khaled Elmously ]
* d/p/lp1947099-honour-user-requested-timeouts-in-all-cases.patch:
Honour user-specified timeouts even in error cases. (LP: #1947099)
[
Setting verification-done-bionic per comments 39/40.
** Tags removed: verification-needed verification-needed-bionic
** Tags added: verification-done verification-done-bionic
--
You received this bug notification because you are a member of STS
Sponsors, which is subscribed to the bug report.
I've also updated the [Test Plan] section of the bug description
** Description changed:
[Impact]
In some cases, ipconfig can take a longer time than the user-specified
timeouts, causing unexpected delays.
[Test Plan]
+
+ - Check that the ipconfig utility is able to obtain an IP
Hello Robie,
I've validated that the package from -proposed works well, testing in my
VM based environment. I haven't tested it on Oracle bare metal (where
the original issue happened) as that is a type of instance hard to get
access to. Given that the test packages had proven to fix the original
I've tested the package from -proposed and I can confirm it fixes the
problem:
Installed from -proposed:
root@ubuntu:~# apt-cache policy klibc-utils
klibc-utils:
Installed: 2.0.4-9ubuntu2.2
Candidate: 2.0.4-9ubuntu2.18.04.1
Version table:
2.0.4-9ubuntu2.18.04.1 500
500
It's not clear to me if upstream have accepted the patch. If not,
https://lists.zytor.com/archives/klibc/2021-December/004635.html sounds
like it's a deliberate upstream design decision not to.
In Ubuntu, we might decide to maintain the patch as a delta but then
drop that delta in subsequent
Hey Fabio,
That happens when the package lands in -proposed.
It looks like the SRU vanguards this Wed/Thu are Robie/Lukasz,
which is good as they are familiar with the discussion/changes,
so I'll ask if they have SRU review cycles to for this upload.
Thanks for following up!
** Changed in:
Should this bug be changed to Fix Committed at this point?
--
You received this bug notification because you are a member of STS
Sponsors, which is subscribed to the bug report.
https://bugs.launchpad.net/bugs/1947099
Title:
ipconfig does not honour user-requested timeouts in some cases
Thanks, Fabio!
With the regression fixed and Oracle test case working as expected,
I've updated the versioning and patch series, and uploaded for SRU.
** Tags added: sts-sponsor-mfo
--
You received this bug notification because you are a member of STS
Sponsors, which is subscribed to the bug
I've tested the new patch from ppa:mfo/lp1947099v2 and I can confirm it
resolves the problem:
- Without the patch:
https://pastebin.ubuntu.com/p/RksNcBGSzn/
It took 396,940865−220,447147 = 176,493718 seconds in the IP-Config section.
Total boot time:
ubuntu@gpu48-ubuntu18:~$ sudo
We'll have to respin/re-upload with a different version, due to the
recent security update [1].
Let's do that after the testing on Oracle, if at all possible, as that
may be asked by the SRU team for accepting it, and some members may be
busy with the jammy release, so let's save one request. :)
Hey, Fabio!
That looks good; thanks!
I have built/uploaded to ppa:mfo/lp1947099v2 and for SRU,
as the tests on Oracle have `-t` in, thus no code change
from previous tests, effectively.
Per our chat, for completeness/correctness you'll test on
oracle again, and then we can move the SRU upload
Thank you, Mauricio, for the build process details and for adding the
update here. I'm including some evidence of my tests showing that the
patch you suggested did work well:
Details of the build process:
https://pastebin.ubuntu.com/p/dmVWH2fxpy/
Test package installed:
Fabio is out today, but he mentioned yesterday that the test patch
worked fine locally, and he'd get access to an oracle test system again,
for more tests.
--
You received this bug notification because you are a member of STS
Sponsors, which is subscribed to the bug report.
Hi Fabio,
> [1] But now when I try to obtain an IP using dhcp without specifying any
> timeout,
> it dumps lots of "Lowered timeout to match user request" messages.
> Is that expected?
>
> https://pastebin.ubuntu.com/p/5Tpc5Rwdkq/
No, this is a regression without `-t`, and has to be fixed
@Łukasz / @Robie, do you think the above comments are enough to proceed
with this SRU?
--
You received this bug notification because you are a member of STS
Sponsors, which is subscribed to the bug report.
https://bugs.launchpad.net/bugs/1947099
Title:
ipconfig does not honour user-requested
I've tested the klibc-utils patch using Mauricio's ppa:
sudo add-apt-repository ppa:mfo/lp1947099
sudo apt install klibc-utils
sudo update-initramfs -u -k all
And I can confirm that it does improve the boot time in more than 3
minutes, without causing any noticeable issues.
- Without the patch:
I tried using klibc-utils from the ppa (containing the patch):
root@ubuntu:~# sudo apt-cache policy klibc-utils
klibc-utils:
Installed: 2.0.4-9ubuntu2.18.04.1
Candidate: 2.0.4-9ubuntu2.18.04.1
Version table:
*** 2.0.4-9ubuntu2.18.04.1 500
500
Hey Fabio,
Thanks for the lab setup for testing this!
> Anyway, if someone can build a klibc-utils package for me with the proposed
> patch,
> I can test the ipconfig behavior in this Lab.
Sure, there you go!
sudo add-apt-repository ppa:mfo/lp1947099
sudo apt install klibc-utils
> For
I've setup a Lab with dnsmasq acting as DHCP Server, which I can use the
dhcp-reply-delay option to introduce a delay between the DHCPDISCOVER
and DHCPOFFER, as in the example below:
Mar 30 18:26:34 focal-dhcpsrv dnsmasq-dhcp[2470]: DHCPDISCOVER(ens3)
52:54:00:d7:10:13
Mar 30 18:26:34
Thank you for the detailed additional information. I think I understand
the situation properly now, together with the need for the fix.
> That means that if something is currently accepting DHCP offers after
the specified timeout has expired, then that is buggy behaviour.
We sometimes run into
A [ Regression potential ] section was added to the description.
@Łukasz Zemczak let me know if there are other details you would like.
Thank you!
** Description changed:
-
[Impact]
In some cases, ipconfig can take a longer time than the user-specified
timeouts, causing unexpected delays.
Hello @Łukasz Zemczak
I understand your concern about breaking existing behaviour.
I would like to clarify 2 things:
1) The change introduced here only ensures that user-requested timeouts are
never exceeded. That means that if something is currently accepting DHCP offers
after the specified
Thank you for the detailed explanation Fabio, the impact of the bug is
now much better understood. Can I ask for an update of the Impact field
in the SRU description above? In a summary, but also getting into a bit
more details about how this differs from focal+ etc.
One thing I worry about in
Hi Robie,
The user story here is about improving the time it takes to boot a
Bionic instance on Oracle Cloud in a specific bare metal shape, called
BM.GPU4.8. This is a pretty large instance, with 18x Ethernet controller
[0200]: Mellanox Technologies MT28800 Family [ConnectX-5 Ex]
[15b3:1019]:
** Changed in: klibc (Ubuntu Bionic)
Status: In Progress => Incomplete
--
You received this bug notification because you are a member of STS
Sponsors, which is subscribed to the bug report.
https://bugs.launchpad.net/bugs/1947099
Title:
ipconfig does not honour user-requested timeouts
> However, Focal (which uses klibc 2.0.7) uses dhclient for networking
initialization instead of ipconfig...
If no users in Focal could be affected by the bug then I'm fine with
accepting the fix to Bionic only. But could you please expand on the
above statement? Under what conditions does Focal
Since the patch has been submitted upstream and there was no hard
objection to it, I think you can proceed in preparing the SRU and
getting it uploaded for SRU review. Just be sure to include the upstream
link in the Forwarded: section of the patch. Thanks!
--
You received this bug notification
Patch proposed upstream:
https://lists.zytor.com/archives/klibc/2021-December/004629.html
--
You received this bug notification because you are a member of STS
Sponsors, which is subscribed to the bug report.
https://bugs.launchpad.net/bugs/1947099
Title:
ipconfig does not honour
Hey guys, sorry for taking so long to comment. I agree with Eric here
that we should get this patch upstreamed. Since we're only seeing this
issue on bionic, I think it's safe to only backport it to bionic - but
preferably after it's submitted upstream and some upstream people
eyeballing it. It's
I'm unsubscribing sponsors given that slashd is handling this.
--
You received this bug notification because you are a member of STS
Sponsors, which is subscribed to the bug report.
https://bugs.launchpad.net/bugs/1947099
Title:
ipconfig does not honour user-requested timeouts in some cases
> I will attempt to upstream this fix to klibc, but I believe the change
to Bionic should happen in parallel/independently since the upstream
patch will not make its way back to Bionic (which is stuck at 2.0.4, as
mentioned above).
Yes that is the plan to SRU the fix on top of what is currently
(updated the bug description with further details according to the SRU
template, including reproduction steps)
** Description changed:
** SRU TEMPLATE DRAFT **
[Impact]
+ In some cases, ipconfig can take a longer time than the user-specified
timeouts, causing unexpected delays.
+
+
Hello @Eric.
> What make Bionic more susceptible to this particular problem ? Bionic
kernel version in use ? else ?
I believe klibc/ipconfig itself is susceptible to the problem in Bionic
and Focal and elsewhere. However, Focal (which uses klibc 2.0.7) uses
dhclient for networking initialization
And let them judge if it's worth an upstream adoption or not as a first
exercise. Then we can take a decision for Ubuntu.
--
You received this bug notification because you are a member of STS
Sponsors, which is subscribed to the bug report.
https://bugs.launchpad.net/bugs/1947099
Title:
I still believe, it would be a good practice to submit the patch to
upstream.
--
You received this bug notification because you are a member of STS
Sponsors, which is subscribed to the bug report.
https://bugs.launchpad.net/bugs/1947099
Title:
ipconfig does not honour user-requested timeouts
And let them judge if it's worth the adoption or not as a first
exercise. Then we can take a decision for Ubuntu.
--
You received this bug notification because you are a member of STS
Sponsors, which is subscribed to the bug report.
https://bugs.launchpad.net/bugs/1947099
Title:
ipconfig does
Khaled El Mously (kmously),
Thanks for the update. I'll review this and talk with sil2100, an SRU
verification member.
Could you please help to fill the SRU template in the description above.
Extra documentations can be found here:
https://wiki.ubuntu.com/StableReleaseUpdates#SRU_Bug_Template
Hi @Eric - thanks for the follow-up and sorry for the delayed reply.
> I don't see this piece of code in the klibc upstream project[0]
Are you referring to the piece of code that I am suggesting as a fix? If
so, it makes sense that it doesn't exist elsewhere, which is why I am
suggesting it be
39 matches
Mail list logo
