| 2019-07-02 01:53:42 |
Felipe Reyes |
description |
I've been analyzing an environment that it's issuing too many times "rndc reload" which eventually leads to the same symptoms seen in bug 1710278 , these are the numbers when the operator attempted to commission a bit more than 50 machines at the same time.
In regiond.log* log files when filtering for a single day (where the test mentioned about above was executed) it was found 5032 events where a new IP was assigned to an existing hostname, and the only 2 hostnames that appear in this list are "ubuntu.maas" and "maas-enlist.maas".
As the DHCP assigns a 600 secs lease time, it's not rare that a single host will renew its IP (and by consequence generate a new event) during the commissioning stage depending on what tests the operator decided to run.
This behavior of MAAS is inducing a high load on "named" which ultimately gets stuck and it stops processing requests and only a restart will take the service back. |
I've been analyzing an environment that it's issuing too many times "rndc reload" which eventually leads to the same symptoms seen in bug 1710278 , these are the numbers when the operator attempted to commission a bit more than 50 machines at the same time.
In regiond.log* log files when filtering for a single day (where the test mentioned about above was executed) it was found 5032 events where a new IP was assigned to an existing hostname, and the only 2 hostnames that appear in this list are "ubuntu.maas" and "maas-enlist.maas".
As the DHCP assigns a 600 secs lease time, it's not rare that a single host will renew its IP (and by consequence generate a new event) during the commissioning stage depending on what tests the operator decided to run.
This behavior of MAAS is inducing a high load on "named" which ultimately gets stuck and it stops processing requests and only a restart will take the service back.
[Environment]
MAAS 2.5.3-7533-g65952b418-0ubuntu1~18.04.1
bind9 1:9.11.3+dfsg-1ubuntu1.8
Ubuntu 18.04 |
|
