2010-08-03 05:16 AM
What i'm trying to achieve is to keep the time diff between filer and domain server not go more than 5min.
how can i do this? i found most of explanations on sync with timeserver a bit confusing.
what could be my optiosn timed value to get not more than 5min between my AD server and timeserver (assuming AD is always in sync with timeserver).
thanks for any help in advance.
2010-08-03 05:21 AM
Just enable timed in FilerView and set it to use NTP as protocol; as servers specify domain controller(s). Make sure to set correct time zone and manually set time close to “real” before (timed won’t correct very large difference).
Something like this:
2010-08-03 01:58 PM
Consider below example:
Filer> options timed
timed.enable on (same value in local+partner recommended)
timed.log off (same value in local+partner recommended)
timed.max_skew 5m (same value in local+partner recommended)
timed.min_skew 0 (same value in local+partner recommended)
timed.proto ntp (same value in local+partner recommended)
timed.sched 1h (same value in local+partner recommended)
timed.servers time.dc.company.net (same value in local+partner recommended)
timed.window 0s (same value in local+partner recommended)
2010-08-04 01:09 PM
Specifies whether time changes initiated by timed should be logged to the console.
Specifies the maximum amount of skew between the time reported by the time server and the filer’s
time that we will allow when synchronizing the time. If the difference in the time reported by the server
and the filer’s time is greater than this value, the filer will not synchronize to the time reported by the
time server. The maximum skew is specified in seconds (suffix s), minutes (suffix m), or hours (suffix
h). Defaults to "30m".
Specifies the protocol used to synchronize time. Valid values for this option are rdate, sntp or rtc.
rdate specifies the rdate (RFC 868) protocol. sntp specifies the Simple Network Time Protocol (RFC
2030). rtc specifies the internal Real-Time Clock chip. The default value for this option is rtc.
Note that ntp is the displayed value for the sntp setting, and can also be used as a valid alias for sntp in
Specifies the timed synchronization schedule. There are several pre-defined schedules:
synchronize every hour (the default)
synchronize every 6 hours
synchronize every day at midnight.
Custom schedules may also be specified by giving the number of minutes or hours between time
synchronization. Minutes are specified by digits followed by an "m"; hours are specified by digits
followed by an "h". For example, options timed.sched 2h will cause time to be synchronized every two
To avoid overburdening the time server, the filer randomly selects the exact time of the synchronization
within a window specified by timed.window.
After timed.sched is set, timed.window is capped at ten percent of timed.sched.
Specifies up to five time servers used by the time daemon. Time servers are contacted in the order
specified; if a server can’t be contacted, the time daemon tries the next one in the list. The default value
for this option is the null string.
Specifies a window around the synchronization time set by timed.sched. The actual synchronization
time is randomly chosen from within this window. timed.window is specified in seconds (suffix s) or
minutes (suffix m). The value may be 0, but it may not exceed ten percent of timed.sched.
timed.window defaults to "0s".
It is recommended that the following list of options have the same value on both filers in a cluster:
2012-05-04 07:47 AM
Thank you Ranil Bhat and Dwaghmare for your info.
If required to change the timed.proto (from rtc to ntp for exapmle) is there any downtime occured or can it be done on the fly? Would any CIFS or NFS shares be affected when changing the protocol type.
I'm just cautious as I want perform the protocol type change on a clustered pair of filers and just hoping there is a 'best practice' type of work-plan someone can guide me to or suggest please?