Comment by martyvis
6 years ago
Only the other week we were doing some testing on a new HCI (hyper-converged infrastructure) I'm doing the network for. At the end of the test period, we were having some storage sync issues. Everything seemed to PING ok, except my colleague happened to notice large jumbo frames over 8000 bytes were getting dropped. I double checked that we hadn't inadvertently changed network configuration. It was only by chance we had another test looking at transceiver signal levels that a customer engineer saw an alarm on RX level. It was then we remembered one test was to remove a module. I then noticed some error counts. We shutdown that particular link until we could visit the site. Sure enough, that fibre wasn't quite clicked in anymore. There was enough of a bridge across the fibre air gap for shall packets, but just wide enough so large packets statically couldn't be corrected enough to work.
Made ne think about the precision required for some errors to occur. Have had sort of similar things occur, and where it's almost impossible to reproduce it when you try!
As a hobbyist sound engineer, usually regular cables are the first I check, but maybe I should extend that to fibre?
Interesting error nevertheless, and honestly, checking fibre cabling for those kind of errors would probably be a bit lower on my list, unless I saw a lot of tranciever errors.