Restarting dnsiq is one of the best practice before performing node reboots
# isi_for_array -s 'ps awux|grep dnsiq_d|grep -v grep'
isi_for_array -s 'ps awux|grep dnsiq_d|grep -v grep'
root 1560 0.0 0.5 54740 7064 ?? Ss 1:03PM 0:00.78 /usr/sbin/isi_dnsiq_d
Check for the status 'Ss' is normal (sleep) - 'Is' was Idle, this would be something to look for.
WCPU @ 0:00.78 is very little, but the process just started, check to see if it differs from the dnsiq daemons from other nodes that do not serve smartconnect requests.
To restart the process (if this was from node 1 - specified with the -x 1 to exclude that node from the command):
# isi_for_array -x 1 killall isi_dnsiq_d
Once the process is killed, it will be automatically started by Master Control Process (MCP).
Once the process is killed, it will be automatically started by Master Control Process (MCP).
This almost worked, DNS had broke on one node and I was incapable of resolving addresses:
ReplyDelete# nslookup
;; connection timed out; trying next origin
;; connection timed out; no servers could be reached
I ended up going to the node on which the dns process did not die and used `kill -9` and the issue was resolved. Thanks for the hints with the DNS service layer, it was not even covered in the Isilon administrators training!
Thanks . This helped.We ran into the same issue.We had a hung DNS causing users to disconnect after a Boot drive replacement on one of the Node . Restarting the process as per the article resolved the issue.
ReplyDelete