convert log time seconds to readable date

[1122633.028643] end_request: I/O error, dev sdc, sector 0

When looking in logs, like dmesg, error messages are preceded by a number that represents the uptime on the server in seconds at the time of the error. So this I/O error happened 1122633 seconds after the machine booted. This means nothing to us. In order to see when the error happened, you need to convert the seconds of uptime into a readable date.

First get the date/time at which the server booted using who -b and convert to seconds. Then add the seconds of uptime from the error message, and then convert back to a human readable date:

# date --date="@$(echo $(date --date="`who -b | awk '{print $3" "$4}'`" +%s)+1122633|bc)"
# Tue Oct 22 00:03:33 PDT 2013

So this error happened shortly after midnight. Very interesting…

Leave a Reply

Your email address will not be published. Required fields are marked *