On Friday 18 November 2005 17:52, Paul Yeatman wrote:
>->>In response to your message<<-
> --received from Gene Heskett--
>
>> On Thursday 17 November 2005 22:02, Jon LaBadie wrote:
>> >On Thu, Nov 17, 2005 at 06:03:08PM -0800, Paul Yeatman wrote:
>> >> Hmmmm, I set the server side etimeout to 2600s. Still get (from
>> >> an amstatus)
>> >>
>> >> pjmac:/ 0 planner: [Estimate timeout from pjmac]
>> >>
>> >> and the amanda report now gives this (mind you, this client
>> >> is the only entry in this "test" config)
>> >>
>> >> Estimate Time (hrs:min) 2:10
>> >>
>> >> 2 hours and 10 minutes!
>> >>
>> >> The message from amandad on the client is virtually the same
>> >>
>> >> <snip>
>> >> amandad: time 0.038: running service
>> >> "/usr/local/libexec/amanda/2.4.5p1/sendsize" amandad: time
>> >> 1244.905: sending REP packet:
>> >> ----
>> >> Amanda 2.4 REP HANDLE 000-30F3BC2A55000000 SEQ 1132261451
>> >> OPTIONS features=fffffeff9ffe7f;
>> >> / 0 SIZE 17648480
>> >> ----
>> >>
>> >> amandad: time 1254.905: dgram_recv: timeout after 10 seconds
>> >> amandad: time 1254.905: waiting for ack: timeout, retrying
>> >> amandad: time 1264.906: dgram_recv: timeout after 10 seconds
>> >> amandad: time 1264.906: waiting for ack: timeout, retrying
>> >> amandad: time 1274.906: dgram_recv: timeout after 10 seconds
>> >> amandad: time 1274.906: waiting for ack: timeout, retrying
>> >> amandad: time 1284.906: dgram_recv: timeout after 10 seconds
>> >> amandad: time 1284.906: waiting for ack: timeout, retrying
>> >> amandad: time 1294.907: dgram_recv: timeout after 10 seconds
>> >> amandad: time 1294.907: waiting for ack: timeout, giving up!
>> >> amandad: time 1294.907: pid 29584 finish time Thu Nov 17 14:52:41
>> >> 2005
>> >>
>> >> Shouldn't the approx. 1255s required to get the estimate fall
>> >> clearly within the 2600s allowed?
>> >>
>> >> Not sure where things are going wrong now
>> >
>> >ISTR someone having a problem with a network setting that caused
>> >connection drops if there was no activity for a particular period.
>> >Could it be something like this, not dealing with amanda at all?
>> >
>> >IIRC the poster was going to try to setup a heartbeat application
>> >running whenever amanda was to try and keep their connection up.
>>
>> I don't recall seeing a message indicating that it worked, so either
>> it did, or that user gave up.
>>
>> Here is what I would do:
>>
>> In roots crontab, setup a ping to the other address over the rf link,
>> using the -i300 option which would ping the other machine at 5 minute
>> intervals. Then, after amdump is done, have it do a 'killall ping'.
>>
>> Or one could use the -c36 number option to have it die after say 36
>> pings which would be 3 hours. Lots of ways to do that & I'm just
>> thinking
>> out loud.
>>
>> Start this about 1 minute before amdump is fired off by the users
>> crontab that runs amdump.
>>
>> And after you have verified that it works, put a redirect to
>> /dev/null in roots crontab entry to kill the email you'll get from
>> it.
>>
>> HTH.
>
>Just to make sure I'm following, the point of this is to make sure that
>the (wireless) internet connection with the laptop is never lost for
>more than 5 minutes?
>
>Paul
No, the point is to make sure it never lost _anytime_ during the backups
"getting it all in one sock" delays, whether its estimate time, or
waiting for tar to make the tarball of a 19Gb tree followed by gzip
-best to smunch a 19 Gb file tar has handed to it.
I suggested 5 minutes, but it might take less to do a full keepalive.
Also a possible consideration is the packet type, it should be a udp,
and without checking the manpages, I don't recall if ping's default
protocol is udp or icmp. So ping might not be the best choice unless
your version of it can be switched with a cli option like traceroute can
be on most platforms. Ping would be the easiest because cli options can
control the interval, and the number of pings, while other methods might
need a short script written to duplicate those functions.
HTH.
--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
99.36% setiathome rank, not too shabby for a WV hillbilly
Yahoo.com and AOL/TW attorneys please note, additions to the above
message by Gene Heskett are:
Copyright 2005 by Maurice Eugene Heskett, all rights reserved.
|
