Bug 32220 - slapd hängt mit 100% CPU
slapd hängt mit 100% CPU
Status: CLOSED FIXED
Product: UCS
Classification: Unclassified
Component: LDAP
UCS 3.1
Other Linux
: P5 normal (vote)
: UCS 3.1-1-errata
Assigned To: Stefan Gohmann
Arvid Requate
http://www.openldap.org/its/index.cgi...
:
Depends on: 20841
Blocks:
  Show dependency treegraph
 
Reported: 2013-08-12 14:42 CEST by Stefan Gohmann
Modified: 2013-08-22 12:06 CEST (History)
6 users (show)

See Also:
What kind of report is it?: ---
What type of bug is this?: ---
Who will be affected by this bug?: ---
How will those affected feel about the bug?: ---
User Pain:
Enterprise Customer affected?:
School Customer affected?:
ISV affected?:
Waiting Support:
Flags outvoted (downgraded) after PO Review:
Ticket number:
Bug group (optional):
Max CVSS v3 score:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Stefan Gohmann univentionstaff 2013-08-12 14:42:43 CEST
We should this fix backport to UCS 3.1-1-errata.

+++ This bug was initially created as a clone of Bug #20841 +++

Auf UCS 2.4-0 und UCS 2.4-1 Master mit 2.6.32-ucs11-amd64
(QA-VM /Univention GmbH/Single/UCS.2-4-0.Master.amd64.tar.gz)
hängt der slapd manchmal mit 100% CPU wenn ein bestimmtest ucs-test-Skript ausgeführt wird.

Reproduzieren:

Folgende Pakete zum Testen installieren
ucs-test-framework          1.3.301-1
ucs-test-libs               1.3.301-1
ucs-test-udm-containers     1.3.301-1

Alle Skripte bis auf 34removecontainerrecursive_ou löschen:
# rm /usr/share/ucs-test/scripts/63_udm-containers/3{{0..3},{5..9}}*

So lange das Testskript ausführen, bis es hängen bleibt (slapd taucht mit 100% CPU in top auf):
# while ucs-test -s udm-containers ; do : ; done
Comment 1 Stefan Gohmann univentionstaff 2013-08-20 08:32:22 CEST
The patch was backported.

3.1-1 Code: r12050
3.1-1 YAML: r43126

OpenLDAP in UCS 3.2 has been updated to a new upstream version which includes the fix: Bug #Bug #20841.
Comment 2 Arvid Requate univentionstaff 2013-08-20 15:19:15 CEST
Verified:

* In ucs3.1-1 the bug was reproducable with the ucs-test case (from 3.2-0) in the 30st test iteration (actually a bit more, running two test loops in parallel).

* After errata3.1-1 update the bug is not reprodicable any more (amd64) within 64+ test iterations (actually quite a bit more, running two test loops in parallel).

* Advisory OK
Comment 3 Janek Walkenhorst univentionstaff 2013-08-22 12:06:58 CEST
http://errata.univention.de/ucs/3.1/169.html