I had blogged about my network problems a few days ago, and now I have dead laptop to add to the fun. This was not really unexpected though. It's been in sad shape for quite some time, and I had been planning to purchase a new one right after Lotusphere. I had put it off for a bit, though, and that's unfortunate because now I'll have to do without until the new Thinkpad T43 I just ordered arrives. The hard drive crashed today. From the sound it's making the chances of recovering data seem pretty slim. Although the laptop itself might be salvageable, it's too old to be worth the effort at this point -- except, possibly, to give to my older daughter.
Despite the fact that the death of the laptop wasn't a huge surprise, I do have to say that today really sucked. Ooops! Rocky says that we shouldn't say things suck any more . But, no. Today sucked hardware-wise.
Last night, I followed up on the prior network/server problems by taking my laptop downstairs, turning off the wireless, setting the wired NIC's speed to 100 Mbit full duplex, enabling it and connecting it directly to my server's NIC with a freshly purchased crossover cable. It worked fine. The Notes client didn't give me any network erorr dialogs at all. I tried another test with wiring both the server and the laptop into a 100 Mbit switch with no other traffic on it, and once again there were no errors. In previous tests with the laptop and server plugged into the same switch, I had gotten errors, but in that test the switch was not dedicated to just those two devices. It had my broadband connection coming in, as well as a desktop, both of which were active during the tests -- plus the upstairs wireless with my wife's laptop, the kids' desktop and the print server, though none of those were generating any activity at the time. That seems to lead to the conclusion that it's not a matter of my server not being able to keep up with the 100 Mbit NIC. Rather, there's something funky going on in the network. It's not the switch, since I tried two different devices and had the same results in the earlier tests. Whatever it is, however, I don't get why it's leading to errors when I use the 100 Mbit switch but isn't leading to errors when I use the 10 Mbit hubs. It remains a mystery.
The thing is, when the test on the isolated switch succeeded, I decided that it was worth another try with the network configured at 100 Mbit across the board, so I did transferred everything downstairs over to the 100 Mbit switch, deactivated the wired port on the laptop and reactivated the wireless. I did some tests from the laptop and everything was fine -- but the wireless access point upstairs was still plugged into a 10 Mbit hub so I picked up the laptop and walked upstairs. I set up a 100 Mbit switch upstairs, moved over all the wires, but when I tried to connect to the server from my laptop I got "Server not responding". After verifying with a ping that I hadn't left a loose wire somewhere, I went downstairs to check the server and it had crashed. Big-time. I brought it back up, and it started doing fixup on a huge database.
Cut to morning. When I get up, I'm told by multiple family members that the server is down. I go downstairs and find that it is up, but not responding at all. In fact, it seems to have locked up a few minutes after I left it the night before. The last messages on the console were from http, indicating that something was trying to pull up a page from a sizeable database that probably needed fixup since it would have most likely been open before last night's first crash. Whether or not fixup was working on it, or it was still waiting I don't know. In any case, I figured that the best course of action would be to (a) revert back to 10 Mbit hubs because a crash and then a hang minutes after I had once again tried 100 Mbit switches is just too suspicious, and (b) I ran fixup separately before bringing up the server.
All was well after that. Until the laptop died.
1. Ken Yee02/13/2006 09:58:33 AM
Sounds like some odd incompatibility w/ the switch upstairs and the one downstairs.
One other thing to try is to disable speed autosensing on the server's NIC. Lock it to 10 or 100. I've seen dropped packets on a network if a NIC sucks. It used to be that 3Com NICs were the safest, but I've been sticking w/ Intel ones lately since their drivers get updated w/ windows update...