cms01 DNS HTTP 10 seconds delay
==================================
.. contents:: :local:
cms01 access monitoring is failing
-------------------------------------
My hourly access/speed test on cms01 began failing about 24hrs ago, sending mails like the below every hour::
cms01.phys.ntu.edu.tw : valmon.py -s envmon : 2013-11-20T09:42:02 : val == 1 : False
val == 1 : False
summary : all:1 True:0 False:1
context : {'val': 0.0}
% ~/.env.cnf blyth@cms01.phys.ntu.edu.tw
[envmon]
tn = envmon
instruction = require a single trunk to be found, verifying that the apache interface to SVN is working
return = int
dbpath = ~/.env/envmon.sqlite
cmd = curl -s --connect-timeout 3 http://dayabay.phys.ntu.edu.tw/repos/env/ | grep trunk | wc -l
note = check C2 server from cron on other nodes
email = blyth@hep1.phys.ntu.edu.tw
constraints = ( val == 1, )
date val
------------------- ----------
2013-11-20T09:42:02 0.0
2013-11-20T08:42:01 0.0
2013-11-20T07:42:01 0.0
2013-11-20T06:42:01 0.0
2013-11-20T05:42:01 0.0
2013-11-20T04:42:01 0.0
2013-11-20T03:42:01 0.0
2013-11-20T02:42:02 0.0
2013-11-20T01:42:01 0.0
2013-11-20T00:42:01 0.0
2013-11-19T23:42:01 0.0
2013-11-19T22:42:01 0.0
2013-11-19T21:42:01 0.0
2013-11-19T20:42:01 0.0
2013-11-19T19:42:01 0.0
2013-11-19T18:42:01 0.0
2013-11-19T17:42:03 0.0
2013-11-19T16:42:02 0.0
2013-11-19T15:42:01 0.0
2013-11-19T14:42:01 0.0
2013-11-19T13:42:02 0.0
2013-11-19T12:42:01 0.0
2013-11-19T11:42:01 0.0
2013-11-19T10:42:01 1.0
Avoiding DNS lookup avoids the delay::
[blyth@cms01 ~]$ time curl http://140.112.101.191/repos/env/ | grep trunk
trunk/
real 0m0.011s
user 0m0.005s
sys 0m0.002s
A reproducible DNS delay of 10 seconds for http GET from cms01 is apparent::
[blyth@cms01 ~]$ time curl http://dayabay.phys.ntu.edu.tw/repos/env/ | grep trunk
trunk/
real 0m10.012s
user 0m0.005s
sys 0m0.004s
[blyth@cms01 ~]$ time curl http://dayabay.phys.ntu.edu.tw/repos/env/ | grep trunk
trunk/
real 0m10.012s
user 0m0.005s
sys 0m0.004s
[blyth@cms01 ~]$
Not just from cms02::
[blyth@cms01 ~]$ time curl http://www.google.com
302 Moved
302 Moved
The document has moved
here.
real 0m10.125s
user 0m0.005s
sys 0m0.003s
[blyth@cms01 ~]$
John sent a mail about DNS recently::
One old DNS server 140.112.101.1 was decommissioned by our dept. system
manager, Debbie. If you are still using this DNS in your machine, you will
encounter some problems. Please change to 140.112.254.4 or 140.112.2.2 or other
DNS servers, ex 168.95.1.1 or 8.8.8.8.
Other Observations
------------------
#. ssh logins to cms01,cms02,hfag have been slow for the past day, several minutes in some cases
DNS /etc/resolv.conf
-----------------------
How to change DNS ?
* :google:`linux change DNS server`
* http://www.rackspace.com/knowledge_center/article/changing-dns-settings-on-linux
* add function ``dns-edit`` to do this
::
[blyth@cms01 ~]$ cat /etc/resolv.conf
search heplocal
search phys.ntu.edu.tw
nameserver 140.112.101.1
nameserver 140.112.2.2
nameserver 168.95.1.1
::
[blyth@cms02 ~]$ cat /etc/resolv.conf
nameserver 140.112.101.1
[blyth@cms02 ~]$ curl http://www.google.com
curl: (6) Couldn't resolve host 'www.google.com'
::
[blyth@hfag blyth]$ cat /etc/resolv.conf
nameserver 140.112.101.1
search phys.ntu.edu.tw
Fixed with ``dns-edit``
-------------------------
#. also note that SSH logins are back to normal, that must be doing some DNS lookups
::
[blyth@cms01 e]$ dns-
[blyth@cms01 e]$ dns-edit
=== dns-edit : sudo vi /etc/resolv.conf
Password:
[blyth@cms01 e]$
[blyth@cms01 e]$
[blyth@cms01 e]$ time curl -s http://www.google.com
302 Moved
302 Moved
The document has moved
here.
real 0m0.051s
user 0m0.003s
sys 0m0.006s
[blyth@cms01 e]$