Bret Wortman wrote:
I think I've found a deeper problem, in that I can't update these
because IPA simply won't start at all now.

I mistyped one of these -- the 2016-03-11 is actually 2018-03-11, and
2016-04-01 is actually 2036-04-01.

As for the unknowns, the first says status: CA_REJECTED and the error
says "hostname in subject of request '' does not match
principal hostname ''", with stuck: yes.

The second is similar, but for a different host.

Is it really a different host and why? I think we'd need to see the full output to know what's going on.

A given host can only get certificates for itself or those delegated to it. Hostnames are used for this enforcement so if they don't line up you'll see this type of rejection.

No idea what's wrong with the rest, or why nothing will start. Near as I
can tell, Kerberos is failing to start, which is causing everything else
to go toes up.

Early in the startup, in /var/log/messages, there's:

ns-slapd: GSSAPI Error: Unspecified GSS failure. Minor code may provide
more information (No Kerberos credentials available)

Without more context it's hard to say. 389 is rather chatty about things and of course when it starts it has no ticket so it logs a bunch of stuff, eventually (hopefully) gets one, and then shuts up.

After that, I get a jar file read pboelm on log4j.jar, then a series of
property setting attempts that don't find matching properties. Then some
cipher errors, then it looks like named starts up okay, and everything
pauses for about 5 minutes before it all comes crashing back down.

I wouldn't get too hung up on particular services just yet. Without valid certs things will fail and those problems will cascade. I think we just need more details at this point.



On 04/26/2016 12:40 PM, Petr Vobornik wrote:
On 04/26/2016 06:00 PM, Bret Wortman wrote:
# getcert list | grep expires
      expires: 2018-04-02 13:04:51 UTC
      expires: 2018-04-02 13:04:31 UTC
      expires: unknown
      expires: 2016-04-17 18:19:19 UTC
      expires: 2016-04-17 18:19:18 UTC
      expires: 2016-04-17 18:19:19 UTC
      expires: 2016-04-01 20:16:39 UTC
      expires: 2016-04-17 18:19:35 UTC
      expires: 2016-03-11 13:04:29 UTC
      expires: unknown

So some got updated and most didn't. Is there a recommended way to update these
all? The system is still backdated to 3 April (ntpd disabled) at this point.
It's usually good to start renewing(when it doesn't happen automatically
from some reason) with the cert which is about to expired first, i.e.
the one with "2016-03-11 13:04:29"

The process is:
- move date before the cert is about to expired
- leave it up to certmonger or manually force resubmit by `getcert
resubmit -i $REQUEST_ID`, where request ID is in `getcert list` output.

I'm little worried about the fact that CA cert was renewed at date which
is after expiration of the other certs.

Also the `expires: unknown` doesn't look good. Check `getcert list`
output for errors related to the cert.


On 04/26/2016 11:46 AM, Petr Vobornik wrote:
On 04/26/2016 03:26 PM, Bret Wortman wrote:
On our non-CA IPA server, this is happening, in case it's related and 

# ipa host-del
ipa: ERROR: Certificate format error: (SEC_ERROR_LEGACY_DATABASE) The
certificate/key database is in an old, unsupported format.
I would start with checking on all IPA servers if and what certificates
are expired:
    # getcert list
or short version to check if there are any:
    # getcert list | grep expires

When CA cert is renewed, it is not automatically transfered to clients.
There one must run:
    # ipa-certupdate

On 04/26/2016 09:24 AM, Bret Wortman wrote:
I rolled the date on the IPA server in question back to April 1 and ran
"ipa-cacert-manage renew", which said it completed successfully. I rolled the
date back to current and tried restarting ipa using ipactl stop && ipactl
start, but no joy. No more ca renewal errors, but right after the pause I see
this in /var/log/messages:

systemd: kadmin.service: main process exited, code=exited,
systemd: Unit kadmin.service entered failed state.
systemd: kadmin.service failed.

I rebooted the server just in case, and it's still getting stuck at the same
place. ipa-otpd doesn't get around to starting.


After the several-minutes-long pause after ipactl start outputs "Starting
pki-tomcatd Service", I get the

On 04/26/2016 08:14 AM, Bret Wortman wrote:
I have an IPA server on a private network which has apparently run into
certificate issues this morning. It's been running without issue for quite a
while, and is on 4.1.4-1 on fedora 21.

This morning, the gui started giving:

IPA Error 907: NetworkError with description "cannot connect to
(SSL_ERROR_EXPIRED_CERRT_ALERT) SSL peer rejected your certificate as expired."

I dug into the logs and after trying to restart ipa using ipactl, there was a
length pause, then:

dogtag-ipa-ca-renew-agent-submit: Updated certificate not available
certmonger: Certificate named "ipaCert" in token "NSS Certificate DB" in
database "/etc/httpd/alias" is no longer valid.
dogtag-ipa-ca-renew-agent-submit: Updated certificate not available
certmonger: Certificate named "ocspSigningCert cert-pki-ca" in token "NSS
Certificate DB" in database "/etc/pki/pki-tomcat/alias" is no longer valid.
dogtag-ipa-ca-renew-agent-submit: Updated certificate not available.
named-pkcs11[3437]: client update
'' denied

and then things start shutting down. I can't start ipa at all using ipactl.

So at present, our DNS is down. Authentication should work for a while, but
I'd like to get this working again as quickly as possible. Any ideas? I deal
with certificates so infrequently (like only when something like this
happens) that I'm not sure where to start.


*Bret Wortman*
/Coming soon to Kickstarter.../

Manage your subscription for the Freeipa-users mailing list:
Go to for more info on the project

Reply via email to