Bug 1899661 - FreeIPA server deployment fails since Fedora-Rawhide-20201119.n.0 with bind "initializing DST: no engine" error
Summary: FreeIPA server deployment fails since Fedora-Rawhide-20201119.n.0 with bind "...
Keywords:
Status: CLOSED RAWHIDE
Alias: None
Product: Fedora
Classification: Fedora
Component: bind-dyndb-ldap
Version: rawhide
Hardware: All
OS: Linux
unspecified
high
Target Milestone: ---
Assignee: Petr Vobornik
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard: openqa
Depends On:
Blocks: F34BetaBlocker
TreeView+ depends on / blocked
 
Reported: 2020-11-19 18:14 UTC by Adam Williamson
Modified: 2020-11-30 17:29 UTC (History)
7 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2020-11-30 17:29:20 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)

Description Adam Williamson 2020-11-19 18:14:04 UTC
With Fedora-Rawhide-20201119.n.0 , the openQA FreeIPA server deployment / upgrade tests all started failing. They all show the same error from bind during startup:

Nov 19 04:05:30 ipa001.domain.local named[33206]: running as: named -u named -c /etc/named.conf -E pkcs11
Nov 19 04:05:30 ipa001.domain.local named[33206]: compiled by GCC 10.2.1 20201016 (Red Hat 10.2.1-6)
Nov 19 04:05:30 ipa001.domain.local named[33206]: compiled with OpenSSL version: OpenSSL 1.1.1g FIPS  21 Apr 2020
Nov 19 04:05:30 ipa001.domain.local named[33206]: linked to OpenSSL version: OpenSSL 1.1.1h FIPS 22 Sep 2020
Nov 19 04:05:30 ipa001.domain.local named[33206]: compiled with libxml2 version: 2.9.10
Nov 19 04:05:30 ipa001.domain.local named[33206]: linked to libxml2 version: 20910
Nov 19 04:05:30 ipa001.domain.local named[33206]: compiled with libjson-c version: 0.14
Nov 19 04:05:30 ipa001.domain.local named[33206]: linked to libjson-c version: 0.14
Nov 19 04:05:30 ipa001.domain.local named[33206]: compiled with zlib version: 1.2.11
Nov 19 04:05:30 ipa001.domain.local named[33206]: linked to zlib version: 1.2.11
Nov 19 04:05:30 ipa001.domain.local named[33206]: threads support is enabled
Nov 19 04:05:30 ipa001.domain.local named[33206]: ----------------------------------------------------
Nov 19 04:05:30 ipa001.domain.local named[33206]: BIND 9 is maintained by Internet Systems Consortium,
Nov 19 04:05:30 ipa001.domain.local named[33206]: Inc. (ISC), a non-profit 501(c)(3) public-benefit
Nov 19 04:05:30 ipa001.domain.local named[33206]: corporation.  Support and training for BIND 9 are
Nov 19 04:05:30 ipa001.domain.local named[33206]: available at https://www.isc.org/support
Nov 19 04:05:30 ipa001.domain.local named[33206]: ----------------------------------------------------
Nov 19 04:05:30 ipa001.domain.local named[33206]: adjusted limit on open files from 524288 to 1048576
Nov 19 04:05:30 ipa001.domain.local named[33206]: found 2 CPUs, using 2 worker threads
Nov 19 04:05:30 ipa001.domain.local named[33206]: using 1 UDP listener per interface
Nov 19 04:05:30 ipa001.domain.local named[33206]: using up to 21000 sockets
Nov 19 04:05:30 ipa001.domain.local named[33206]: initializing DST: no engine
Nov 19 04:05:30 ipa001.domain.local named[33206]: exiting (due to fatal error)
Nov 19 04:05:30 ipa001.domain.local systemd[1]: named.service: Control process exited, code=exited, status=1/FAILURE
Nov 19 04:05:30 ipa001.domain.local systemd[1]: named.service: Failed with result 'exit-code'.

this results in ipa.service failing (as part of initial deployment in the deployment tests, and on first boot after upgrade in the upgrade tests).

There was a bumper crop of FreeIPA-related updates in this compose:

Package:      bind-dyndb-ldap-11.5-1.fc34
Old package:  bind-dyndb-ldap-11.3-5.fc34

Package:      freeipa-4.9.0-0.rc1.fc34
Old package:  freeipa-4.8.10-7.fc34

Package:      krb5-1.18.3-2.fc34
Old package:  krb5-1.18.2-30.fc34

Package:      openldap-2.4.56-1.fc34
Old package:  openldap-2.4.55-1.fc34

Package:      python-ldap-3.3.1-2.fc34
Old package:  python-ldap-3.3.1-1.fc34

Package:      tomcat-1:9.0.40-1.fc34
Old package:  tomcat-1:9.0.39-1.fc34

...filing against bind-dyndb-ldap to start with as a guess. Proposing as a Beta blocker as a violation of Basic criterion https://fedoraproject.org/wiki/Basic_Release_Criteria#FreeIPA_server_requirements - "It must be possible to configure a Fedora Server system installed according to the above criteria as a FreeIPA domain controller, using the official deployment tools provided in the distribution FreeIPA packages."

Comment 1 Adam Williamson 2020-11-19 18:35:59 UTC
Oh, bad me, I made an assumption that turns out wrong - the upgrade tests aren't failing on exactly this, though they're still failing on a bind problem. bind crashes on startup after the upgrade, then the clients can't resolve names. I'll file a separate bug for that crash.

Comment 2 Alexander Bokovoy 2020-11-19 18:46:04 UTC
I think this is due to a spec changes I did in freeipa.spec.

In particular, this line is breaking %{with bind_pkcs11} logic:
https://src.fedoraproject.org/rpms/freeipa/blob/master/f/freeipa.spec#_115

I addressed that in https://github.com/freeipa/freeipa/pull/5279/files#diff-79e7e776c34748018cf388f4492c4b28a4212e1ed49dfd826c34d370106233d1L110-L115 but it is not yet merged as we haven't yet completed the unification of the spec files.

I'm doing a build now.

Comment 3 Alexander Bokovoy 2020-11-19 19:11:48 UTC
https://koji.fedoraproject.org/koji/taskinfo?taskID=55896678 should address this issue.

Comment 4 Adam Williamson 2020-11-19 21:43:26 UTC
Filed https://bugzilla.redhat.com/show_bug.cgi?id=1899744 for the bind crash on upgrade.

Comment 5 Alexander Bokovoy 2020-11-27 09:13:06 UTC
With bug 1899744 fixed with bind-dyndb-ldap 11.6-1.fc34, and python3-dns downgraded to Fedora 33 version (bug 1902061), I get successful deployment of IPA master and replica on Rawhide.

Comment 6 Adam Williamson 2020-11-30 17:29:20 UTC
Well, in openQA tests we seem to be still failing in named startup. Different error, though, and it happens slightly later than this one did:

Nov 30 05:14:09 ipa001.domain.local named[33077]: unable to open directory 'dyndb-ldap', working directory is '/var/named': permission denied
Nov 30 05:14:09 ipa001.domain.local named[33077]: LDAP config validation failed for database 'ipa': permission denied
Nov 30 05:14:09 ipa001.domain.local named[33077]: dynamic database 'ipa' configuration failed: permission denied
Nov 30 05:14:09 ipa001.domain.local named[33077]: loading configuration: permission denied
Nov 30 05:14:09 ipa001.domain.local named[33077]: exiting (due to fatal error)
Nov 30 05:14:09 ipa001.domain.local systemd[1]: named.service: Control process exited, code=exited, status=1/FAILURE
Nov 30 05:14:09 ipa001.domain.local systemd[1]: named.service: Failed with result 'exit-code'.
Nov 30 05:14:09 ipa001.domain.local systemd[1]: Failed to start Berkeley Internet Name Domain (DNS).

so I think we can say this one is fixed, and I'll file a new bug.


Note You need to log in before you can comment on or make changes to this bug.