On 06-11-2024 13:59, Rob Crittenden wrote:
Kees Bakker via FreeIPA-users wrote:
On 04-11-2024 12:56, Florence Blanc-Renaud wrote:
Hi,
On Fri, Nov 1, 2024 at 3:51 PM Kees Bakker via FreeIPA-users
<[email protected]> wrote:
Hi,
After a scheduled power down on and power up on one of our IPA
servers
named crashed.
I remember I saw this before and that it was "solved" by
restarting named.
This time however it took about 12 retries before named finally
continued.
This is the error it produces
../../../lib/dns/name.c:667: REQUIRE((name1->attributes &
0x00000001) ==
(name2->attributes & 0x00000001)) failed, back trace
bind9 is quite drastic when it hits a failing REQUIRE. It immediately
stops with a coredump.
But it doesn't tell which names are failing the comparison.
Of course I can (or should) report this to the bind developers.
However,
I'm sure they
are going to say that this is an old version and that I should ask
CentOS/RH devs.
The issue has already been reported here:
https://issues.redhat.com/browse/RHEL-30407
If anybody here on the list has an idea, then please let me know.
Some more details
Version info:
* os: CentOS 9-Stream
* ipa packages 4.12.0-4.el9
* bind9 packages 32:9.16.23-15.el9
* bind-libdyndb-ldap 11.9.8-el9
Some updates are available for those packages, it may be worth trying.
Nope. With the latest CentOS 9-Stream packages it is also crashing at
startup.
Version info:
* ipa packages: 4.12.2-1.el9
* bind packages: 32:9.16.23-24.el9
* bind-libdyndb-ldap: 11.9-10.el9
The only thing for me to do is to keep retrying to start named until it
finally stays up and running.
It would help if I knew a little bit more what is in that RHEL-30407 issue.
It probably wouldn't. There are some stack traces there with a similar
assertion failure but no solutions yet. If you want to provide a stack
trace we could confirm it is identical but in the end it would just be
"yup, same thing" and still wouldn't be that helpful.
The bind developer has been looking into it but the root cause has been
elusive.
rob
When you said "bind developer", did you mean someone from RedHat, or
someone from the Bind ISC community?
The reason I ask is that I have now a debug environment setup where I can
run gdb and break at the point where it hits the assertion_failed.
The problem is, I have to make sense of what I'm looking at. I want to get
advice from a bind developer how to print information, for example,
(gdb) up
#3 0x00007ffff7cb4adc in match (name2=0x7fffd8650010,
name1=0x7fff8ccf5010) at ../../../lib/dns/order.c:96
96 return (dns_name_equal(name1, name2));
(gdb) p *name1
$1 = {magic = 1145983854, ndata = 0x7fff8ccec010 "", length = 0, labels
= 0, attributes = 0, offsets = 0x7fff8ccf5068 "",
buffer = 0x0, link = {prev = 0x0, next = 0x0}, list = {head =
0x7fff8cb71c50, tail = 0x7fff8cb71c50}, ht = 0x0}
How can I see the actual name from this? That's the question I would ask.
--
Kees
--
_______________________________________________
FreeIPA-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
Fedora Code of Conduct:
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives:
https://lists.fedorahosted.org/archives/list/[email protected]
Do not reply to spam, report it:
https://pagure.io/fedora-infrastructure/new_issue
