EON: NTP problems

[Travis T]

I've been using EON successfully (mostly) for several months now.

Occasionally, I notice my shares are not accessible from my windows boxes. I typically see some smb error in the /var/adm/messages that is not too specific. The typical fix is to rejoin the domain using smbadm join.

Today, I've been fighting to get this working again all day. I finally got it, but I don't expect it to last long. It started with one of my nfs shares not being accessible from my VMWare host. Because my DCs run on VMware, EON couldn't contact my PDC, and I couldn't browse my shares from my windows boxes. After getting the nfs share back online and my DCs powered back up, there were different errors in the messages file.

I started getting clock skew too great errors from ntp. I configured the ntp.conf to look to my DCs for time. I googled how to restart the ntp service, but nothing worked with EON. I compared the clock on my DC to the clock on EON and it was maybe a minute or two off. After running ntpdate -u <PDC>, the times synced up and I stopped getting that error.

Ended up having more problems and had to populate the krb5.conf file (not sure how this was working before with just defaults), and when I was done, I got another clock skew too great. ntpdate -u fixed it and I could successfully join the domain and browse shares.

How can I fix this permenantly?

[Andre Lue]

Hi Travist,

"clock skew too great" are usually related to the clock on the client and the AD server being too far apart (5mins if I remember correctly). Time being synced is fairly critical with these kinds of setup.

If EON is being synced from this same AD server, the only thing I can say is to make sure both servers are using the same timezone (eg PST8PDT for both).

You can also check the drift file for the difference. Other than that it seems something else is causing the time to be off or drifting apart.

see the command set to help trouble shoot and correspond the values of time
http://sites.google.com/site/eonstorage/faq#TOC-How-do-I-check-internet-time-sync-

[Travis T]

Dre,

I was glad to see that error, because that (possibly) explains the intermittent problems I've been having. I thought it would be an easy fix, but because there were default time servers loaded into the ntp.conf file previously, ntpq -c peers still shows them even though I have updated the conf file to show only my DCs. I can force an sync from my DC, but then the clock starts drifting back to the time source ntp is syncing to.

Bottom line, is there a way to restart ntp to force a re-read of the ntp.conf file without rebooting?

[Andre Lue]

edit the /etc/inet/ntp.conf with the new time server entries. replace the 4 default, with your own
server 192.43.244.18 # time.nist.gov
server 207.171.7.151 # pool.ntp.org
server 216.184.20.83 # pool.ntp.org
server 129.6.15.29 # time-b.nist.gov

stop the ntpd
pkill ntpd

run that cmd to start ntp (grep ntp /mnt/eon0/.exec to find the line)
/usr/lib/inet/ntpd -p /var/run/ntp.pid -g

that should get you all sync'd against your time server