pcp icon indicating copy to clipboard operation
pcp copied to clipboard

Ubuntu 22.04 support

Open jasonk000 opened this issue 3 years ago • 10 comments

opening as a tracking log / current status,

building here.

jasonk000 avatar Jun 03 '22 17:06 jasonk000

first pass on my local machine resulted in a few failures

(294 326 532 918 1661).
Failed 5 of 1305 tests

It was good to see it build without issues so I'll spin it up on a full size instance and proceed from here.

jasonk000 avatar Jun 03 '22 17:06 jasonk000

G'day @jasonk000

I've upgraded one of my local Ubuntu systems to 22.04 ... but I'm seeing different QA failures:

1385 1701 1702 1704 1706 1707 1722 1723 1724 1725

while all of the failures mentioned in your post above are passing for me. Are yours "hard" or "flakey" failures? You can use the check-flakey script in the qa dir to rerun tests with .out.bad files and this will give some indication if the failures are flakey or not. If you have "hard" failures, perhaps you could email me the .out.bad and .full files for the failing tests.

My failures are (a) the openmetrics PMDA goes nuts (1385) and (b) bpftrace(1) looks broken or a victim of the LTO change (all of the 17?? ones). Did check run these tests for you? It is possible you don't have optional packages installed or services running, so these ones were "notrun" for you ... it would be nice to know if any of my failures passed (as opposed to notrun) on your system.

Also, over on slack you mentioned some "package changes" for Ubuntu 22.04 ... if these are not already in

qa/admin/package-lists/Ubuntu+22.04+x86_64

I'd be interested in the details.

Cheers, Ken.

kmcdonell avatar Jun 07 '22 01:06 kmcdonell

A few extra packages required to support the pmdabpf implementation (since libbpf is available now on ubuntu22).

From an ec2 instance with a built pcp-6.x trunk, I see the following errors.

Failures: 036 063 119 151 169 241 244 250 255 279 280 294 304 338 344 359 392 429 430 457 471 503 529 530 532 641 649 658 664 674 681 686 805 870 892 893 925 999 1049 1087 1108 1166 1190 1297 1351 1426 1429 1480 1546 1701 1722 1723 1768 1897

I've not yet reviewed them though, but only a couple of overlaps.

jasonk000 avatar Jul 30 '22 01:07 jasonk000

I'll keep a list running list of issues found / resolved

  • Additional packages etc were required, to install fakeroot and the bpf PMDA dependencies. I put these in the ubuntu-2204 branch.
  • sudo apt-get update ; sudo apt-get -y install build-essential
  • sudo mount -o remount,exec /var/tmp required as /var/tmp is by default mounted as noexec
  • sudo mount -o remount,hidepid=0 /proc required since our installation hardening breaks things -- alternatively, sudo usermod -a -G nac pcpqa ; sudo usermod -a -G nac pcp ; sudo usermod -a -G pcp pcpqa but this doesn't seem to work
  • git clone --depth 1 --branch ubuntu-2204 https://github.com/jasonk000/pcp.git ; cd pcp ; ./qa/admin/list-packages -x cpan | xargs sudo apt-get -y install
  • vim Makepkgs add export MAKEFLAGS=-j32, and also unset MAKEFLAGS before the debian packaging calls (inside debian_buildpackage())
  • ./Makepkgs --verbose ; sudo dpkg -i build/deb/*.deb
  • echo 'pcpqa ALL=(ALL) NOPASSWD: ALL' | sudo tee /etc/sudoers.d/pcpqa ; sudo usermod -a -G pcp pcpqa
  • sudo chkconfig pmlogger on ; sudo systemctl start pmlogger ; sudo systemctl start pmcd
  • cd /var/lib/pcp/testsuite ; sudo -u pcpqa ./check 000

fails

  • 430 => fails because it assumes one IP, however our /etc/hosts has two (ipv4, ipv6)
  • 892 => fails because it looks for an IPv6 localhost, but our /etc/hosts does not have it - fix by adding ip6-localhost to the ::1 line
  • 1480 => fails because our sensor output is giving temp1_input: -273.150 on this instance
  • 1701, 1722, 1723 => the test script fails due to upstream debian bpftrace issue, more in https://github.com/ajor/bpftrace/issues/50

failed first run, passed second

  • 608 749

tbd

  • 886 results mismatch in indom kernel.cpu.interrupts

jasonk000 avatar Aug 01 '22 22:08 jasonk000

I've added 1701, 1722 and 1723 to the triaged list for Ubuntu 22.04 pending the arrival of the upstream fix from https://github.com/iovisor/bpftrace/pull/2264

kmcdonell avatar Aug 16 '22 23:08 kmcdonell

@jasonk000 if 608 and/or 749 fail again could you please either email the .out.bad and .fail files to me, or attach 'em to this issue?

Ditto for 886.

kmcdonell avatar Aug 16 '22 23:08 kmcdonell

qa/430 is fixed in commit 045e5d8b4

kmcdonell avatar Aug 17 '22 07:08 kmcdonell

qa/1480 is fixed in commit 0d56ed200

kmcdonell avatar Aug 17 '22 07:08 kmcdonell

qa/892 is now _notrun if no IPv6 entry for localhost in /etc/hosts. Fix is in commit 8a9da013d

kmcdonell avatar Aug 17 '22 07:08 kmcdonell

qa/749 was not being run here (systemtap was not in the packages list). When I installed that the test passes. The error ./scripts/pahole-flags.sh: line 7: return: can only `return' from a function or sourced script has nothing to do with PCP, it comes from linux-headers. I really don't want to filter it away in qa/749, so would rather leave it failing until linux-headers is fixed.

kmcdonell avatar Aug 20 '22 03:08 kmcdonell

I think this can be closed now, correct? Anything outstanding? @kmcdonell @natoscott

jasonk000 avatar Oct 28 '22 15:10 jasonk000

All sorted I believe, yep.

natoscott avatar Oct 29 '22 01:10 natoscott

Oh, I should also mention - you can follow daily CI status on all platforms via the #pcpqa channel on PCP slack.

natoscott avatar Oct 29 '22 01:10 natoscott

Hi Jason,

I don't have any outstanding issues specific to this version of Ubuntu.  I've been using Ubuntu 22.04 for some time as one of my development systems and PCP seems just fine there.

Cheers, Ken.

On 29/10/2022 2:26 am, Jason Koch wrote:

I think this can be closed now, correct? Anything outstanding? @kmcdonell https://github.com/kmcdonell @natoscott https://github.com/natoscott

— Reply to this email directly, view it on GitHub https://github.com/performancecopilot/pcp/issues/1609#issuecomment-1295134863, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABDJ7KY5NNC4P3RC6IRIRPLWFPWCPANCNFSM5XZWM5JQ. You are receiving this because you were mentioned.Message ID: @.***>

kmcdonell avatar Oct 29 '22 02:10 kmcdonell