[posted to rush.general]
Hi Luigi.
Out of interest are you seeing these memory exceptions on Linux render
nodes? I am more used to seeing that kind of thing with windows.
How about passing the log file through a script that adds date to
the start
of each line. Some clever awk'ing and tee'ing could do I should think.
|-----Original Message-----
|From: Luigi Tommaseo [mailto:luigi@tommaseo.me.uk]
|Sent: Monday, 23 January 2006 9:26 PM
|To: void@(email surpressed)
|Subject: Re: timestamp on logs
|
|[posted to rush.general]
|
|well what we would like to have is the frame log with a time
|stamp for each line to compare the time our renders throw
|memory exceptions with the network spikes that we are seeing
|to be able to see if there is a relation between the 2.
|at the moment it looks as if converting maya to mi throws
|memory exception even though the ram is available. not sure
|about what it is I would like to be able to rule out the
|server performance.
|maybe some sort of logging of the server and the frame log
|related to each other would help.
|
|thanks
|
|Luigi Tommaseo
|Digital Systems Manager
|Senate Visual Effects
|Twickenham Film Studios
|St.Margarets
|Twickenham
|Middlesex
|TW1 2AW
|
|Tel: (Tel# suppressed)
|luigi@(email surpressed)
|www.senatevfx.com
|
|On 20 Jan 2006, at 19:32, Greg Ercolano wrote:
|
|> [posted to rush.general]
|>
|> Luigi Tommaseo wrote:
|>> is there any way we can have time stamp on the logs? what we would
|>> like to do here is to create a network and server
|performance log and
|>> correlate it to the logs of the render boxes to see if the
problems
|>> we are having in the conversion to mentalray are indeed related to
|>> network performance or something else.
|>
|> Hi Luigi,
|>
|> Which logs, the frame logs, or the cpu accounting logs?
|>
|> There are several time stamps in the frame logs:
|>
|> o The file's date stamp on the log itself
|> o The file's 'Started:' date stamp from the log,
ie:
|>
|> % grep Started: /path/to/log/0001
|> -- Started: Thu Jan 05 14:46:14 2006
|>
|> In the cpu accounting logs (rush/var/cpu.acct), there's
|date stamps
|> on all the entries in time(2) format.
|>
|> You should be able to come up with a simple script to dive into
|> all the cpu.acct logs, sort the entries by date, and then easily
|> be able to cross reference the frames that rendered vs.
|the errors
|> in your server performance log.
|>
|> I'd go for the cpu.acct logs on the remotes; those are easy to
|> collect and sort, and has all the job
|title/owner/frame#/exit code
|> info in them.
|>
|> You might also cross reference errors with your system logs on
|> each render machine, ie. /var/log/messages (LINUX),
|> /var/log/system.log (OSX), or the windows event manager logs.
|>
|> --
|> Greg Ercolano, erco@(email surpressed)
|> Rush Render Queue, http://seriss.com/rush/
|> Tel: xxx-xxx-xxxx
|> Cel: xxx-xxx-xxxx
|> Fax: xxx-xxx-xxx
|