Hello.
This is a general question regarding troubleshooting approach.
Suppose there is a problem like this: some mainframe computer tries to transfer 1000 files to a SQL server and occasionally a database application displays error saying that there was some kind of failure with network and some part of those 1000 files is not delivered. This has been going on for a week or so. SQL people are thus convinced that there's some malfunction within the network infrastructure and asks network people to investigate.
Let's say that there are 12 hops between source and destination and a networking guy must do initial troubleshooting using only command line commands, wthout any third party network analyzing/monitoring tools (and without using debugging).
What commands whoud you use to determine if there is actually something wrong with network?
What I would use:
1. Traceroute from source to destination to check latency.
2. Ping from source to destination to check packet loss.
3. Check the interfaces directly connected to source and destination devices and see if there is increasing count of errors. Also check if there are some logs regarding the interfaces.
Anything else to check for an initial troubleshooting? As I mentioned, there are 12 hops between source and destination - would you login to each of those 12 hops and check all involved interfaces if they contain some errors and also check logs?
I do realize that if SQL application throws some error with the phrase "network failure", it does not necessary mean that the problem exist somewhere in networking devices and not in the SQL server itself. But in this case I still need to convince non-network people that from the networks side everything is OK.
Thank you.
