lots of software updater requests to the webserver in a small time period. about softwareupdater HOT 3 CLOSED

Author: couvb
These nodes seem to all be from the undergrad labs. I suspect something strange about the lab setup is causing them to generate the same random number sequences (possibly due to synchronized times).

To further investigate this, I will deploy a simple script to all the lab computers that use a random wait scheme similar to the one in the software updater, except it will send debugging information to a server that I control. This will hopefully provide more insight into what exactly is going on, and how we might deal with it.

from softwareupdater.

Author: ivan
You can read the apache's access logs on seattle.cs here:
/var/log/apache2/access.log

To grab just those that hit the updatesite (via the software updater) do:
$ cat /var/log/apache2/access.log | grep updatesite | less

You'll find a couple of timestamps at which the nodes are synchronized in hitting the webserver. For example, from the current log file here is a list of top 50 timestamps (just year, and timestamp) along with the total number of requests that occurred on the exact timestamp in the log. Note that the timestamp "07:50:01" has been the most popular, but not on the same day as in the actual log you won't find any day where more than about 15 hosts make requests at exactly this time.

2009:07:50:01 102
2009:08:50:01 98
2009:07:10:01 98
2009:07:20:01 97
2009:09:00:01 96
2009:08:10:01 95
2009:17:00:02 94
2009:08:20:01 94
2009:06:50:01 94
2009:08:40:01 93
2009:08:00:01 93
2009:07:40:01 93
2009:07:30:01 93
2009:10:10:01 92
2009:06:40:01 92
2009:09:20:01 91
2009:09:30:01 90
2009:10:00:01 89
2009:13:10:02 88
2009:09:50:01 88
2009:08:30:01 88
2009:14:00:02 86
2009:13:00:02 86
2009:07:00:01 85
2009:16:50:02 83
2009:13:50:01 83
2009:12:10:01 83
2009:17:40:01 82
2009:14:20:02 82
2009:12:50:02 82
2009:10:30:01 82
2009:10:20:01 82
2009:09:10:01 82
2009:12:20:01 81
2009:11:50:01 81
2009:11:40:01 81
2009:17:50:01 80
2009:14:30:02 80
2009:11:20:01 80
2009:11:00:01 80
2009:10:50:01 80
2009:17:10:02 77
2009:16:20:01 77
2009:12:40:02 77
2009:12:30:02 77
2009:15:20:01 76
2009:13:40:02 76
2009:11:10:01 76
2009:16:40:02 75
2009:10:40:01 75

from softwareupdater.

Author: ivan
This seems to have resolved itself. We probably also need some way of monitoring this over a long time period.

from softwareupdater.