On 05/29/2017 08:24 AM, Sigbjorn Lie-Soland via FreeIPA-users wrote:
I've had several discussions, tickets, crash dumps, logs, etc with the
team over the years regarding this issue. The issue disappeared on
rhel 6 after the troubleshooting, however it's re-appeared on rhel 7.
This is by no means a permanent fix. It's a workaround. The mentioned
troubleshooting needs to be done as well. However having dns
unavailbility is a critical issue, and as there is no potential db to
recover after a named crash, this should be a safe workaround.
Den 27. mai 2017 03:34:37 CEST, skrev John Keates via FreeIPA-users
At the risk of smelling like a thread hijack; I’m experiencing the same issue on one
server (Fedora 25), but on all others it’s fine. I don’t think this is a ‘normal’ issue
that should be ‘fixed’ by restarting named-pkcs11 all the time.
I tend to check for known issues (and solutions) on this issue from time to time, but
there doesn’t seem to be a standard issue with a standard solution for this case (yet?).
I suppose the FreeIPA team would like a ticket, some logs, maybe some core dumps or
gdb outputs, but I’d create my own message thread or ticket and not bother you here ;-)
On 26 May 2017, at 19:11, Sigbjorn Lie via FreeIPA-users
<freeipa-users(a)lists.fedorahosted.org> wrote: Hi, I have
experienced named stopping unexpectedly from time to time.
After moving to RHEL 7 the I made use of a handy feature in
systemd, “Restart=always”, to make sure named is kept alive.
This has kept named alive for me, and I was wondering if this
perhaps would be a useful addition to the default
"named-pkcs11.service” shipped in RHEL? The changes I have
done is to copy the file
/etc/systemd/system/named-pkcs11.service, and adding the
following to [service] section: --- Restart=always
RestartSec=3 — The underlying issue of why named is crashing
would of course also needs to be investigated separately. What
do you think? Regards, Siggi
I've recently examined this bug, but I wasn't able to find any issue on
bind-dyndb-ldap side. The crashes might be caused by bind itself.
Bug for reference: https://bugzilla.redhat.com/show_bug.cgi?id=1435270
Restarting the service is possible, so I think it's a reasonable
workaround to restart it in case it crashes.
PGP: 4A8B A48C 2AED 933B D495 C509 A1FB A5F7 EF8C 4869