Something weird and unresolved for today:
So this automated process gets server host name, and sents it as part of a HTTP POST to another server. The data was encrypted with SSL in transmission. When it arrived, it was changed from:
jrepublic.somedomain.com
to
jrepbulic.somedomain.com
I ran it again manually, and it behaved normally, without the byteswap. Now, the way to generate this error, in software is something like this:
xchg al,ah
But there is nothing like that …
The source system:
- CPU: Intel(R) Core(TM)2 Quad CPU Q8400 @ 2.66GHz
- Motherboard: Intel DG43GT
- Temperature: 40 degrees C
- Ethernet controller: 8086:10ce: Intel Corporation 82567V-2 Gigabit Network Connection
- Client: python 2.4.3
The target system:
- CPU: Intel(R) Core(TM)2 CPU 4400 @ 2.00GHz
- Motherboard: Intel DG33BU
- Ethernet controller: 8086:294c: Intel Corporation Gigabit Ethernet Controller (rev 02)
- Temperature: 42 degrees C
- Server: python 2.4.3
The traffic traversed the following systems:
- 4 Cisco switches
- Fortigate firewall (two of these)
- Cisco backbone stuff
- Possibly 1 Netgear switch
I suspect a CPU error on the originating server, since it is likely that SSL would stop if there was an error … but really don’t know. A byteswap on the network that just happened to pass the checksumming in SSL is not beyond belief. Oh well.
