On Friday, April 15, 2011 12:39:19 PM Tim Johnson did opine: > I have already tried changing the etimeout, as stated before the > timeouts are : > etimeout 8000 (changed from 5000 and the origanal -600) > ( the long wait is due to a few of the machines being very old 800 MHz) > dtimeout 2800 > ctimeout 60 > > I believe I tried using a -5000 seconds and see if it worked > before, however I will try it now and see what happens. > > On Fri, 15 Apr 2011, gene heskett wrote: > > On Friday, April 15, 2011 10:17:50 AM Tim Johnson did opine: > >> Just a follow up on this issue, which isnt makeing > >> > >> sense to me at all. > >> > >> I shut down the amanda client via xinetd, and ran > >> > >> the backups and the backups worked perfectly. It went over > >> the client with the report mentioning that the client > >> was not responsive. > >> > >> However, I had someone (by accident?) turn off the power > >> > >> to one of the amanda clients. This caused the backup on the > >> server to continue to wait for the estimate of the client > >> for hours (over 10 hours). > >> > >> I have tried using calcsize, client, and no entry for > >> > >> the estimate (I dont know what the default is, calcsize (?) ). > >> > >> No error messages in the logs, just the hanging server waiting > >> > >> for the estimate of the non responsive client. > >> > >> I will be trying different options to see what occurs. > >> > >> Thanks for any input. > > > > That is controlled by the 'etimeout' value in your amanda.conf, and > > its stated in seconds. IIRC I'm set at 800 seconds but that is > > overkill by at least a factor of 2. That is about 13 minutes in > > round figures. Depending on one drive speeds and iron in the cpu, 300 > > (5 minutes) is probably adequate for most smaller uses. > > > > [...]
At 8000, that would be 2:22 hours, per missing DLE, although if on different 'spindles' this can be done in parallel by amanda. I would not set it at more than 200% of the longest etime consumed during a normal, all DLE's present backup. I believe each physical drive in the system should have its own unique 'spindle' number in the disklist in order for this to work as intended. It can make a noticeable diff in overall backup time even if there are only 2 'spindle's being backed up. I am guessing that having >10 would probably need a gigabyte network to stay within the available bandwidth though. However, if you are approaching that scenario, it would be a good selling point to TPTB that you need a network upgrade if yours is only 100 megabaud now. -- 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) <http://tinyurl.com/ddg5bz> <http://www.cantrip.org/gatto.html> There are twenty-five people left in the world, and twenty-seven of them are hamburgers. -- Ed Sanders
